Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _LINUX_FS_H |
2 | #define _LINUX_FS_H | |
3 | ||
1da177e4 LT |
4 | |
5 | #include <linux/linkage.h> | |
6 | #include <linux/wait.h> | |
1da177e4 LT |
7 | #include <linux/kdev_t.h> |
8 | #include <linux/dcache.h> | |
3f8206d4 | 9 | #include <linux/path.h> |
1da177e4 LT |
10 | #include <linux/stat.h> |
11 | #include <linux/cache.h> | |
1da177e4 | 12 | #include <linux/list.h> |
bc3b14cb | 13 | #include <linux/list_lru.h> |
4f5e65a1 | 14 | #include <linux/llist.h> |
1da177e4 | 15 | #include <linux/radix-tree.h> |
6b2dbba8 | 16 | #include <linux/rbtree.h> |
1da177e4 | 17 | #include <linux/init.h> |
914e2637 | 18 | #include <linux/pid.h> |
187f1882 | 19 | #include <linux/bug.h> |
1b1dcc1b | 20 | #include <linux/mutex.h> |
c8c06efa | 21 | #include <linux/rwsem.h> |
3bd858ab | 22 | #include <linux/capability.h> |
6188e10d | 23 | #include <linux/semaphore.h> |
c4b929b8 | 24 | #include <linux/fiemap.h> |
ceb5bdc2 | 25 | #include <linux/rculist_bl.h> |
07b8ce1e | 26 | #include <linux/atomic.h> |
83aeeada | 27 | #include <linux/shrinker.h> |
c1aab02d | 28 | #include <linux/migrate_mode.h> |
92361636 | 29 | #include <linux/uidgid.h> |
5accdf82 | 30 | #include <linux/lockdep.h> |
c2b1ad80 | 31 | #include <linux/percpu-rwsem.h> |
08cce05c | 32 | #include <linux/blk_types.h> |
1da177e4 | 33 | |
1da177e4 | 34 | #include <asm/byteorder.h> |
607ca46e | 35 | #include <uapi/linux/fs.h> |
1da177e4 | 36 | |
b83ae6d4 | 37 | struct backing_dev_info; |
a5694255 | 38 | struct export_operations; |
a885c8c4 | 39 | struct hd_geometry; |
1da177e4 | 40 | struct iovec; |
92198f7e | 41 | struct kiocb; |
57cc7215 | 42 | struct kobject; |
1da177e4 LT |
43 | struct pipe_inode_info; |
44 | struct poll_table_struct; | |
45 | struct kstatfs; | |
46 | struct vm_area_struct; | |
47 | struct vfsmount; | |
745ca247 | 48 | struct cred; |
a509bc1a | 49 | struct swap_info_struct; |
55985dd7 | 50 | struct seq_file; |
7b7a8665 | 51 | struct workqueue_struct; |
3b93f911 | 52 | struct iov_iter; |
f7ca90b1 | 53 | struct vm_fault; |
1da177e4 | 54 | |
74bf17cf | 55 | extern void __init inode_init(void); |
1da177e4 | 56 | extern void __init inode_init_early(void); |
1da177e4 LT |
57 | extern void __init files_init(unsigned long); |
58 | ||
dded4f4d | 59 | extern struct files_stat_struct files_stat; |
518de9b3 | 60 | extern unsigned long get_max_files(void); |
dded4f4d JE |
61 | extern int sysctl_nr_open; |
62 | extern struct inodes_stat_t inodes_stat; | |
63 | extern int leases_enable, lease_break_time; | |
800179c9 KC |
64 | extern int sysctl_protected_symlinks; |
65 | extern int sysctl_protected_hardlinks; | |
dded4f4d | 66 | |
1da177e4 LT |
67 | struct buffer_head; |
68 | typedef int (get_block_t)(struct inode *inode, sector_t iblock, | |
69 | struct buffer_head *bh_result, int create); | |
92198f7e | 70 | typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset, |
7b7a8665 | 71 | ssize_t bytes, void *private); |
1da177e4 | 72 | |
bbc1096a DH |
73 | #define MAY_EXEC 0x00000001 |
74 | #define MAY_WRITE 0x00000002 | |
75 | #define MAY_READ 0x00000004 | |
76 | #define MAY_APPEND 0x00000008 | |
77 | #define MAY_ACCESS 0x00000010 | |
78 | #define MAY_OPEN 0x00000020 | |
79 | #define MAY_CHDIR 0x00000040 | |
80 | /* called from RCU mode, don't block */ | |
81 | #define MAY_NOT_BLOCK 0x00000080 | |
82 | ||
83 | /* | |
84 | * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond | |
85 | * to O_WRONLY and O_RDWR via the strange trick in __dentry_open() | |
86 | */ | |
87 | ||
88 | /* file is open for reading */ | |
89 | #define FMODE_READ ((__force fmode_t)0x1) | |
90 | /* file is open for writing */ | |
91 | #define FMODE_WRITE ((__force fmode_t)0x2) | |
92 | /* file is seekable */ | |
93 | #define FMODE_LSEEK ((__force fmode_t)0x4) | |
94 | /* file can be accessed using pread */ | |
95 | #define FMODE_PREAD ((__force fmode_t)0x8) | |
96 | /* file can be accessed using pwrite */ | |
97 | #define FMODE_PWRITE ((__force fmode_t)0x10) | |
98 | /* File is opened for execution with sys_execve / sys_uselib */ | |
99 | #define FMODE_EXEC ((__force fmode_t)0x20) | |
100 | /* File is opened with O_NDELAY (only set for block devices) */ | |
101 | #define FMODE_NDELAY ((__force fmode_t)0x40) | |
102 | /* File is opened with O_EXCL (only set for block devices) */ | |
103 | #define FMODE_EXCL ((__force fmode_t)0x80) | |
104 | /* File is opened using open(.., 3, ..) and is writeable only for ioctls | |
105 | (specialy hack for floppy.c) */ | |
106 | #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100) | |
107 | /* 32bit hashes as llseek() offset (for directories) */ | |
108 | #define FMODE_32BITHASH ((__force fmode_t)0x200) | |
109 | /* 64bit hashes as llseek() offset (for directories) */ | |
110 | #define FMODE_64BITHASH ((__force fmode_t)0x400) | |
111 | ||
112 | /* | |
113 | * Don't update ctime and mtime. | |
114 | * | |
115 | * Currently a special hack for the XFS open_by_handle ioctl, but we'll | |
116 | * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. | |
117 | */ | |
118 | #define FMODE_NOCMTIME ((__force fmode_t)0x800) | |
119 | ||
120 | /* Expect random access pattern */ | |
121 | #define FMODE_RANDOM ((__force fmode_t)0x1000) | |
122 | ||
123 | /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */ | |
124 | #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000) | |
125 | ||
126 | /* File is opened with O_PATH; almost nothing can be done with it */ | |
127 | #define FMODE_PATH ((__force fmode_t)0x4000) | |
128 | ||
9c225f26 LT |
129 | /* File needs atomic accesses to f_pos */ |
130 | #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) | |
83f936c7 AV |
131 | /* Write access to underlying fs */ |
132 | #define FMODE_WRITER ((__force fmode_t)0x10000) | |
7f7f25e8 AV |
133 | /* Has read method(s) */ |
134 | #define FMODE_CAN_READ ((__force fmode_t)0x20000) | |
135 | /* Has write method(s) */ | |
136 | #define FMODE_CAN_WRITE ((__force fmode_t)0x40000) | |
9c225f26 | 137 | |
bbc1096a | 138 | /* File was opened by fanotify and shouldn't generate fanotify events */ |
75069f2b | 139 | #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) |
bbc1096a DH |
140 | |
141 | /* | |
142 | * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector | |
143 | * that indicates that they should check the contents of the iovec are | |
144 | * valid, but not check the memory that the iovec elements | |
145 | * points too. | |
146 | */ | |
147 | #define CHECK_IOVEC_ONLY -1 | |
148 | ||
08cce05c DH |
149 | /* |
150 | * The below are the various read and write types that we support. Some of | |
151 | * them include behavioral modifiers that send information down to the | |
152 | * block layer and IO scheduler. Terminology: | |
153 | * | |
154 | * The block layer uses device plugging to defer IO a little bit, in | |
155 | * the hope that we will see more IO very shortly. This increases | |
156 | * coalescing of adjacent IO and thus reduces the number of IOs we | |
157 | * have to send to the device. It also allows for better queuing, | |
158 | * if the IO isn't mergeable. If the caller is going to be waiting | |
159 | * for the IO, then he must ensure that the device is unplugged so | |
160 | * that the IO is dispatched to the driver. | |
161 | * | |
162 | * All IO is handled async in Linux. This is fine for background | |
163 | * writes, but for reads or writes that someone waits for completion | |
164 | * on, we want to notify the block layer and IO scheduler so that they | |
165 | * know about it. That allows them to make better scheduling | |
166 | * decisions. So when the below references 'sync' and 'async', it | |
167 | * is referencing this priority hint. | |
168 | * | |
169 | * With that in mind, the available types are: | |
170 | * | |
171 | * READ A normal read operation. Device will be plugged. | |
172 | * READ_SYNC A synchronous read. Device is not plugged, caller can | |
173 | * immediately wait on this read without caring about | |
174 | * unplugging. | |
175 | * READA Used for read-ahead operations. Lower priority, and the | |
176 | * block layer could (in theory) choose to ignore this | |
177 | * request if it runs into resource problems. | |
178 | * WRITE A normal async write. Device will be plugged. | |
179 | * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down | |
180 | * the hint that someone will be waiting on this IO | |
181 | * shortly. The write equivalent of READ_SYNC. | |
182 | * WRITE_ODIRECT Special case write for O_DIRECT only. | |
183 | * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush. | |
184 | * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on | |
185 | * non-volatile media on completion. | |
186 | * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded | |
187 | * by a cache flush and data is guaranteed to be on | |
188 | * non-volatile media on completion. | |
189 | * | |
190 | */ | |
191 | #define RW_MASK REQ_WRITE | |
192 | #define RWA_MASK REQ_RAHEAD | |
193 | ||
194 | #define READ 0 | |
195 | #define WRITE RW_MASK | |
196 | #define READA RWA_MASK | |
08cce05c DH |
197 | |
198 | #define READ_SYNC (READ | REQ_SYNC) | |
199 | #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE) | |
200 | #define WRITE_ODIRECT (WRITE | REQ_SYNC) | |
201 | #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH) | |
202 | #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA) | |
203 | #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA) | |
204 | ||
1da177e4 LT |
205 | /* |
206 | * Attribute flags. These should be or-ed together to figure out what | |
207 | * has been changed! | |
208 | */ | |
9767d749 MS |
209 | #define ATTR_MODE (1 << 0) |
210 | #define ATTR_UID (1 << 1) | |
211 | #define ATTR_GID (1 << 2) | |
212 | #define ATTR_SIZE (1 << 3) | |
213 | #define ATTR_ATIME (1 << 4) | |
214 | #define ATTR_MTIME (1 << 5) | |
215 | #define ATTR_CTIME (1 << 6) | |
216 | #define ATTR_ATIME_SET (1 << 7) | |
217 | #define ATTR_MTIME_SET (1 << 8) | |
218 | #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ | |
219 | #define ATTR_ATTR_FLAG (1 << 10) | |
220 | #define ATTR_KILL_SUID (1 << 11) | |
221 | #define ATTR_KILL_SGID (1 << 12) | |
222 | #define ATTR_FILE (1 << 13) | |
223 | #define ATTR_KILL_PRIV (1 << 14) | |
224 | #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ | |
225 | #define ATTR_TIMES_SET (1 << 16) | |
1da177e4 | 226 | |
787fb6bc MS |
227 | /* |
228 | * Whiteout is represented by a char device. The following constants define the | |
229 | * mode and device number to use. | |
230 | */ | |
231 | #define WHITEOUT_MODE 0 | |
232 | #define WHITEOUT_DEV 0 | |
233 | ||
1da177e4 LT |
234 | /* |
235 | * This is the Inode Attributes structure, used for notify_change(). It | |
236 | * uses the above definitions as flags, to know which values have changed. | |
237 | * Also, in this manner, a Filesystem can look at only the values it cares | |
238 | * about. Basically, these are the attributes that the VFS layer can | |
239 | * request to change from the FS layer. | |
240 | * | |
241 | * Derek Atkins <warlord@MIT.EDU> 94-10-20 | |
242 | */ | |
243 | struct iattr { | |
244 | unsigned int ia_valid; | |
245 | umode_t ia_mode; | |
92361636 EB |
246 | kuid_t ia_uid; |
247 | kgid_t ia_gid; | |
1da177e4 LT |
248 | loff_t ia_size; |
249 | struct timespec ia_atime; | |
250 | struct timespec ia_mtime; | |
251 | struct timespec ia_ctime; | |
cc4e69de MS |
252 | |
253 | /* | |
25985edc | 254 | * Not an attribute, but an auxiliary info for filesystems wanting to |
cc4e69de MS |
255 | * implement an ftruncate() like method. NOTE: filesystem should |
256 | * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). | |
257 | */ | |
258 | struct file *ia_file; | |
1da177e4 LT |
259 | }; |
260 | ||
1da177e4 LT |
261 | /* |
262 | * Includes for diskquotas. | |
263 | */ | |
264 | #include <linux/quota.h> | |
265 | ||
69c433ed MS |
266 | /* |
267 | * Maximum number of layers of fs stack. Needs to be limited to | |
268 | * prevent kernel stack overflow | |
269 | */ | |
270 | #define FILESYSTEM_MAX_STACK_DEPTH 2 | |
271 | ||
994fc28c ZB |
272 | /** |
273 | * enum positive_aop_returns - aop return codes with specific semantics | |
274 | * | |
275 | * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has | |
276 | * completed, that the page is still locked, and | |
277 | * should be considered active. The VM uses this hint | |
278 | * to return the page to the active list -- it won't | |
279 | * be a candidate for writeback again in the near | |
280 | * future. Other callers must be careful to unlock | |
281 | * the page if they get this return. Returned by | |
282 | * writepage(); | |
283 | * | |
284 | * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has | |
285 | * unlocked it and the page might have been truncated. | |
286 | * The caller should back up to acquiring a new page and | |
287 | * trying again. The aop will be taking reasonable | |
288 | * precautions not to livelock. If the caller held a page | |
289 | * reference, it should drop it before retrying. Returned | |
55144768 | 290 | * by readpage(). |
994fc28c ZB |
291 | * |
292 | * address_space_operation functions return these large constants to indicate | |
293 | * special semantics to the caller. These are much larger than the bytes in a | |
294 | * page to allow for functions that return the number of bytes operated on in a | |
295 | * given page. | |
296 | */ | |
297 | ||
298 | enum positive_aop_returns { | |
299 | AOP_WRITEPAGE_ACTIVATE = 0x80000, | |
300 | AOP_TRUNCATED_PAGE = 0x80001, | |
301 | }; | |
302 | ||
afddba49 | 303 | #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */ |
89e10787 | 304 | #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */ |
54566b2c NP |
305 | #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct |
306 | * helper code (eg buffer layer) | |
307 | * to clear GFP_FS from alloc */ | |
afddba49 | 308 | |
1da177e4 LT |
309 | /* |
310 | * oh the beauties of C type declarations. | |
311 | */ | |
312 | struct page; | |
313 | struct address_space; | |
314 | struct writeback_control; | |
1da177e4 | 315 | |
e2e40f2c | 316 | #define IOCB_EVENTFD (1 << 0) |
2ba48ce5 AV |
317 | #define IOCB_APPEND (1 << 1) |
318 | #define IOCB_DIRECT (1 << 2) | |
e2e40f2c CH |
319 | |
320 | struct kiocb { | |
321 | struct file *ki_filp; | |
322 | loff_t ki_pos; | |
323 | void (*ki_complete)(struct kiocb *iocb, long ret, long ret2); | |
324 | void *private; | |
325 | int ki_flags; | |
326 | }; | |
327 | ||
328 | static inline bool is_sync_kiocb(struct kiocb *kiocb) | |
329 | { | |
330 | return kiocb->ki_complete == NULL; | |
331 | } | |
332 | ||
2ba48ce5 AV |
333 | static inline int iocb_flags(struct file *file); |
334 | ||
e2e40f2c CH |
335 | static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) |
336 | { | |
337 | *kiocb = (struct kiocb) { | |
338 | .ki_filp = filp, | |
2ba48ce5 | 339 | .ki_flags = iocb_flags(filp), |
e2e40f2c CH |
340 | }; |
341 | } | |
342 | ||
8ab22b9a HH |
343 | /* |
344 | * "descriptor" for what we're up to with a read. | |
345 | * This allows us to use the same read code yet | |
346 | * have multiple different users of the data that | |
347 | * we read from a file. | |
348 | * | |
349 | * The simplest case just copies the data to user | |
350 | * mode. | |
351 | */ | |
352 | typedef struct { | |
353 | size_t written; | |
354 | size_t count; | |
355 | union { | |
356 | char __user *buf; | |
357 | void *data; | |
358 | } arg; | |
359 | int error; | |
360 | } read_descriptor_t; | |
361 | ||
362 | typedef int (*read_actor_t)(read_descriptor_t *, struct page *, | |
363 | unsigned long, unsigned long); | |
2f718ffc | 364 | |
1da177e4 LT |
365 | struct address_space_operations { |
366 | int (*writepage)(struct page *page, struct writeback_control *wbc); | |
367 | int (*readpage)(struct file *, struct page *); | |
1da177e4 LT |
368 | |
369 | /* Write back some dirty pages from this mapping. */ | |
370 | int (*writepages)(struct address_space *, struct writeback_control *); | |
371 | ||
4741c9fd | 372 | /* Set a page dirty. Return true if this dirtied it */ |
1da177e4 LT |
373 | int (*set_page_dirty)(struct page *page); |
374 | ||
375 | int (*readpages)(struct file *filp, struct address_space *mapping, | |
376 | struct list_head *pages, unsigned nr_pages); | |
377 | ||
afddba49 NP |
378 | int (*write_begin)(struct file *, struct address_space *mapping, |
379 | loff_t pos, unsigned len, unsigned flags, | |
380 | struct page **pagep, void **fsdata); | |
381 | int (*write_end)(struct file *, struct address_space *mapping, | |
382 | loff_t pos, unsigned len, unsigned copied, | |
383 | struct page *page, void *fsdata); | |
384 | ||
1da177e4 LT |
385 | /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ |
386 | sector_t (*bmap)(struct address_space *, sector_t); | |
d47992f8 | 387 | void (*invalidatepage) (struct page *, unsigned int, unsigned int); |
27496a8c | 388 | int (*releasepage) (struct page *, gfp_t); |
6072d13c | 389 | void (*freepage)(struct page *); |
22c6186e | 390 | ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset); |
b969c4ab | 391 | /* |
ef277c73 TC |
392 | * migrate the contents of a page to the specified target. If |
393 | * migrate_mode is MIGRATE_ASYNC, it must not block. | |
b969c4ab | 394 | */ |
2d1db3b1 | 395 | int (*migratepage) (struct address_space *, |
a6bc32b8 | 396 | struct page *, struct page *, enum migrate_mode); |
e3db7691 | 397 | int (*launder_page) (struct page *); |
c186afb4 | 398 | int (*is_partially_uptodate) (struct page *, unsigned long, |
8ab22b9a | 399 | unsigned long); |
b4597226 | 400 | void (*is_dirty_writeback) (struct page *, bool *, bool *); |
25718736 | 401 | int (*error_remove_page)(struct address_space *, struct page *); |
62c230bc MG |
402 | |
403 | /* swapfile support */ | |
a509bc1a MG |
404 | int (*swap_activate)(struct swap_info_struct *sis, struct file *file, |
405 | sector_t *span); | |
406 | void (*swap_deactivate)(struct file *file); | |
1da177e4 LT |
407 | }; |
408 | ||
7dcda1c9 JA |
409 | extern const struct address_space_operations empty_aops; |
410 | ||
afddba49 NP |
411 | /* |
412 | * pagecache_write_begin/pagecache_write_end must be used by general code | |
413 | * to write into the pagecache. | |
414 | */ | |
415 | int pagecache_write_begin(struct file *, struct address_space *mapping, | |
416 | loff_t pos, unsigned len, unsigned flags, | |
417 | struct page **pagep, void **fsdata); | |
418 | ||
419 | int pagecache_write_end(struct file *, struct address_space *mapping, | |
420 | loff_t pos, unsigned len, unsigned copied, | |
421 | struct page *page, void *fsdata); | |
422 | ||
1da177e4 LT |
423 | struct address_space { |
424 | struct inode *host; /* owner: inode, block_device */ | |
425 | struct radix_tree_root page_tree; /* radix tree of all pages */ | |
19fd6231 | 426 | spinlock_t tree_lock; /* and lock protecting it */ |
4bb5f5d9 | 427 | atomic_t i_mmap_writable;/* count VM_SHARED mappings */ |
6b2dbba8 | 428 | struct rb_root i_mmap; /* tree of private and shared mappings */ |
c8c06efa | 429 | struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */ |
08142579 | 430 | /* Protected by tree_lock together with the radix tree */ |
1da177e4 | 431 | unsigned long nrpages; /* number of total pages */ |
91b0abe3 | 432 | unsigned long nrshadows; /* number of shadow entries */ |
1da177e4 | 433 | pgoff_t writeback_index;/* writeback starts here */ |
f5e54d6e | 434 | const struct address_space_operations *a_ops; /* methods */ |
1da177e4 | 435 | unsigned long flags; /* error bits/gfp mask */ |
1da177e4 LT |
436 | spinlock_t private_lock; /* for use by the address_space */ |
437 | struct list_head private_list; /* ditto */ | |
252aa6f5 | 438 | void *private_data; /* ditto */ |
1da177e4 LT |
439 | } __attribute__((aligned(sizeof(long)))); |
440 | /* | |
441 | * On most architectures that alignment is already the case; but | |
25985edc | 442 | * must be enforced here for CRIS, to let the least significant bit |
1da177e4 LT |
443 | * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. |
444 | */ | |
87192a2a | 445 | struct request_queue; |
1da177e4 LT |
446 | |
447 | struct block_device { | |
448 | dev_t bd_dev; /* not a kdev_t - it's a search key */ | |
f3ccfcda | 449 | int bd_openers; |
1da177e4 | 450 | struct inode * bd_inode; /* will die */ |
87d8fe1e | 451 | struct super_block * bd_super; |
c039e313 | 452 | struct mutex bd_mutex; /* open/close mutex */ |
1da177e4 | 453 | struct list_head bd_inodes; |
6b4517a7 | 454 | void * bd_claiming; |
1da177e4 LT |
455 | void * bd_holder; |
456 | int bd_holders; | |
77ea887e | 457 | bool bd_write_holder; |
641dc636 | 458 | #ifdef CONFIG_SYSFS |
49731baa | 459 | struct list_head bd_holder_disks; |
641dc636 | 460 | #endif |
1da177e4 LT |
461 | struct block_device * bd_contains; |
462 | unsigned bd_block_size; | |
463 | struct hd_struct * bd_part; | |
464 | /* number of times partitions within this device have been opened. */ | |
465 | unsigned bd_part_count; | |
466 | int bd_invalidated; | |
467 | struct gendisk * bd_disk; | |
87192a2a | 468 | struct request_queue * bd_queue; |
1da177e4 | 469 | struct list_head bd_list; |
1da177e4 LT |
470 | /* |
471 | * Private data. You must have bd_claim'ed the block_device | |
472 | * to use this. NOTE: bd_claim allows an owner to claim | |
473 | * the same device multiple times, the owner must take special | |
474 | * care to not mess up bd_private for that case. | |
475 | */ | |
476 | unsigned long bd_private; | |
fcccf502 TS |
477 | |
478 | /* The counter of freeze processes */ | |
479 | int bd_fsfreeze_count; | |
480 | /* Mutex for freeze */ | |
481 | struct mutex bd_fsfreeze_mutex; | |
1da177e4 LT |
482 | }; |
483 | ||
484 | /* | |
485 | * Radix-tree tags, for tagging dirty and writeback pages within the pagecache | |
486 | * radix trees | |
487 | */ | |
488 | #define PAGECACHE_TAG_DIRTY 0 | |
489 | #define PAGECACHE_TAG_WRITEBACK 1 | |
f446daae | 490 | #define PAGECACHE_TAG_TOWRITE 2 |
1da177e4 LT |
491 | |
492 | int mapping_tagged(struct address_space *mapping, int tag); | |
493 | ||
8b28f621 DB |
494 | static inline void i_mmap_lock_write(struct address_space *mapping) |
495 | { | |
c8c06efa | 496 | down_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
497 | } |
498 | ||
499 | static inline void i_mmap_unlock_write(struct address_space *mapping) | |
500 | { | |
c8c06efa | 501 | up_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
502 | } |
503 | ||
3dec0ba0 DB |
504 | static inline void i_mmap_lock_read(struct address_space *mapping) |
505 | { | |
506 | down_read(&mapping->i_mmap_rwsem); | |
507 | } | |
508 | ||
509 | static inline void i_mmap_unlock_read(struct address_space *mapping) | |
510 | { | |
511 | up_read(&mapping->i_mmap_rwsem); | |
512 | } | |
513 | ||
1da177e4 LT |
514 | /* |
515 | * Might pages of this file be mapped into userspace? | |
516 | */ | |
517 | static inline int mapping_mapped(struct address_space *mapping) | |
518 | { | |
27ba0644 | 519 | return !RB_EMPTY_ROOT(&mapping->i_mmap); |
1da177e4 LT |
520 | } |
521 | ||
522 | /* | |
523 | * Might pages of this file have been modified in userspace? | |
524 | * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff | |
525 | * marks vma as VM_SHARED if it is shared, and the file was opened for | |
526 | * writing i.e. vma may be mprotected writable even if now readonly. | |
4bb5f5d9 DH |
527 | * |
528 | * If i_mmap_writable is negative, no new writable mappings are allowed. You | |
529 | * can only deny writable mappings, if none exists right now. | |
1da177e4 LT |
530 | */ |
531 | static inline int mapping_writably_mapped(struct address_space *mapping) | |
532 | { | |
4bb5f5d9 DH |
533 | return atomic_read(&mapping->i_mmap_writable) > 0; |
534 | } | |
535 | ||
536 | static inline int mapping_map_writable(struct address_space *mapping) | |
537 | { | |
538 | return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? | |
539 | 0 : -EPERM; | |
540 | } | |
541 | ||
542 | static inline void mapping_unmap_writable(struct address_space *mapping) | |
543 | { | |
544 | atomic_dec(&mapping->i_mmap_writable); | |
545 | } | |
546 | ||
547 | static inline int mapping_deny_writable(struct address_space *mapping) | |
548 | { | |
549 | return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? | |
550 | 0 : -EBUSY; | |
551 | } | |
552 | ||
553 | static inline void mapping_allow_writable(struct address_space *mapping) | |
554 | { | |
555 | atomic_inc(&mapping->i_mmap_writable); | |
1da177e4 LT |
556 | } |
557 | ||
558 | /* | |
559 | * Use sequence counter to get consistent i_size on 32-bit processors. | |
560 | */ | |
561 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
562 | #include <linux/seqlock.h> | |
563 | #define __NEED_I_SIZE_ORDERED | |
564 | #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) | |
565 | #else | |
566 | #define i_size_ordered_init(inode) do { } while (0) | |
567 | #endif | |
568 | ||
f19d4a8f AV |
569 | struct posix_acl; |
570 | #define ACL_NOT_CACHED ((void *)(-1)) | |
571 | ||
3ddcd056 LT |
572 | #define IOP_FASTPERM 0x0001 |
573 | #define IOP_LOOKUP 0x0002 | |
574 | #define IOP_NOFOLLOW 0x0004 | |
575 | ||
576 | /* | |
577 | * Keep mostly read-only and often accessed (especially for | |
578 | * the RCU path lookup and 'stat' data) fields at the beginning | |
579 | * of the 'struct inode' | |
580 | */ | |
1da177e4 | 581 | struct inode { |
44a7d7a8 | 582 | umode_t i_mode; |
3ddcd056 | 583 | unsigned short i_opflags; |
92361636 EB |
584 | kuid_t i_uid; |
585 | kgid_t i_gid; | |
3ddcd056 LT |
586 | unsigned int i_flags; |
587 | ||
588 | #ifdef CONFIG_FS_POSIX_ACL | |
589 | struct posix_acl *i_acl; | |
590 | struct posix_acl *i_default_acl; | |
591 | #endif | |
592 | ||
44a7d7a8 NP |
593 | const struct inode_operations *i_op; |
594 | struct super_block *i_sb; | |
3ddcd056 | 595 | struct address_space *i_mapping; |
44a7d7a8 | 596 | |
13e12d14 LT |
597 | #ifdef CONFIG_SECURITY |
598 | void *i_security; | |
599 | #endif | |
44a7d7a8 | 600 | |
3ddcd056 LT |
601 | /* Stat data, not accessed from path walking */ |
602 | unsigned long i_ino; | |
a78ef704 MS |
603 | /* |
604 | * Filesystems may only read i_nlink directly. They shall use the | |
605 | * following functions for modification: | |
606 | * | |
607 | * (set|clear|inc|drop)_nlink | |
608 | * inode_(inc|dec)_link_count | |
609 | */ | |
610 | union { | |
611 | const unsigned int i_nlink; | |
612 | unsigned int __i_nlink; | |
613 | }; | |
3ddcd056 | 614 | dev_t i_rdev; |
2f9d3df8 | 615 | loff_t i_size; |
3ddcd056 LT |
616 | struct timespec i_atime; |
617 | struct timespec i_mtime; | |
618 | struct timespec i_ctime; | |
6cdbb0ef TT |
619 | spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ |
620 | unsigned short i_bytes; | |
2f9d3df8 | 621 | unsigned int i_blkbits; |
3ddcd056 LT |
622 | blkcnt_t i_blocks; |
623 | ||
624 | #ifdef __NEED_I_SIZE_ORDERED | |
625 | seqcount_t i_size_seqcount; | |
626 | #endif | |
627 | ||
628 | /* Misc */ | |
629 | unsigned long i_state; | |
3ddcd056 | 630 | struct mutex i_mutex; |
13e12d14 | 631 | |
44a7d7a8 | 632 | unsigned long dirtied_when; /* jiffies of first dirtying */ |
a2f48706 | 633 | unsigned long dirtied_time_when; |
44a7d7a8 | 634 | |
1da177e4 | 635 | struct hlist_node i_hash; |
7ccf19a8 NP |
636 | struct list_head i_wb_list; /* backing dev IO list */ |
637 | struct list_head i_lru; /* inode LRU list */ | |
1da177e4 | 638 | struct list_head i_sb_list; |
fa0d7e3d | 639 | union { |
b3d9b7a3 | 640 | struct hlist_head i_dentry; |
fa0d7e3d NP |
641 | struct rcu_head i_rcu; |
642 | }; | |
7a224228 | 643 | u64 i_version; |
2f9d3df8 | 644 | atomic_t i_count; |
bd5fe6c5 | 645 | atomic_t i_dio_count; |
6cdbb0ef | 646 | atomic_t i_writecount; |
d984ea60 MZ |
647 | #ifdef CONFIG_IMA |
648 | atomic_t i_readcount; /* struct files open RO */ | |
649 | #endif | |
99ac48f5 | 650 | const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ |
4a075e39 | 651 | struct file_lock_context *i_flctx; |
1da177e4 | 652 | struct address_space i_data; |
1da177e4 | 653 | struct list_head i_devices; |
4c154168 TT |
654 | union { |
655 | struct pipe_inode_info *i_pipe; | |
eaf796e7 | 656 | struct block_device *i_bdev; |
577c4eb0 | 657 | struct cdev *i_cdev; |
61ba64fc | 658 | char *i_link; |
4c154168 | 659 | }; |
1da177e4 LT |
660 | |
661 | __u32 i_generation; | |
662 | ||
3be25f49 EP |
663 | #ifdef CONFIG_FSNOTIFY |
664 | __u32 i_fsnotify_mask; /* all events this inode cares about */ | |
e61ce867 | 665 | struct hlist_head i_fsnotify_marks; |
0eeca283 RL |
666 | #endif |
667 | ||
8e18e294 | 668 | void *i_private; /* fs or device private pointer */ |
1da177e4 LT |
669 | }; |
670 | ||
1d3382cb AV |
671 | static inline int inode_unhashed(struct inode *inode) |
672 | { | |
673 | return hlist_unhashed(&inode->i_hash); | |
674 | } | |
675 | ||
f2eace23 IM |
676 | /* |
677 | * inode->i_mutex nesting subclasses for the lock validator: | |
678 | * | |
679 | * 0: the object of the current VFS operation | |
680 | * 1: parent | |
681 | * 2: child/target | |
40bd22c9 BF |
682 | * 3: xattr |
683 | * 4: second non-directory | |
d1b72cc6 MS |
684 | * 5: second parent (when locking independent directories in rename) |
685 | * | |
686 | * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two | |
40bd22c9 | 687 | * non-directories at once. |
f2eace23 IM |
688 | * |
689 | * The locking order between these classes is | |
d1b72cc6 | 690 | * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory |
f2eace23 IM |
691 | */ |
692 | enum inode_i_mutex_lock_class | |
693 | { | |
694 | I_MUTEX_NORMAL, | |
695 | I_MUTEX_PARENT, | |
696 | I_MUTEX_CHILD, | |
4df46240 | 697 | I_MUTEX_XATTR, |
d1b72cc6 MS |
698 | I_MUTEX_NONDIR2, |
699 | I_MUTEX_PARENT2, | |
f2eace23 IM |
700 | }; |
701 | ||
375e289e BF |
702 | void lock_two_nondirectories(struct inode *, struct inode*); |
703 | void unlock_two_nondirectories(struct inode *, struct inode*); | |
704 | ||
1da177e4 LT |
705 | /* |
706 | * NOTE: in a 32bit arch with a preemptable kernel and | |
707 | * an UP compile the i_size_read/write must be atomic | |
708 | * with respect to the local cpu (unlike with preempt disabled), | |
709 | * but they don't need to be atomic with respect to other cpus like in | |
710 | * true SMP (so they need either to either locally disable irq around | |
711 | * the read or for example on x86 they can be still implemented as a | |
712 | * cmpxchg8b without the need of the lock prefix). For SMP compiles | |
713 | * and 64bit archs it makes no difference if preempt is enabled or not. | |
714 | */ | |
48ed214d | 715 | static inline loff_t i_size_read(const struct inode *inode) |
1da177e4 LT |
716 | { |
717 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
718 | loff_t i_size; | |
719 | unsigned int seq; | |
720 | ||
721 | do { | |
722 | seq = read_seqcount_begin(&inode->i_size_seqcount); | |
723 | i_size = inode->i_size; | |
724 | } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); | |
725 | return i_size; | |
726 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) | |
727 | loff_t i_size; | |
728 | ||
729 | preempt_disable(); | |
730 | i_size = inode->i_size; | |
731 | preempt_enable(); | |
732 | return i_size; | |
733 | #else | |
734 | return inode->i_size; | |
735 | #endif | |
736 | } | |
737 | ||
7762f5a0 MS |
738 | /* |
739 | * NOTE: unlike i_size_read(), i_size_write() does need locking around it | |
740 | * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount | |
741 | * can be lost, resulting in subsequent i_size_read() calls spinning forever. | |
742 | */ | |
1da177e4 LT |
743 | static inline void i_size_write(struct inode *inode, loff_t i_size) |
744 | { | |
745 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
74e3d1e1 | 746 | preempt_disable(); |
1da177e4 LT |
747 | write_seqcount_begin(&inode->i_size_seqcount); |
748 | inode->i_size = i_size; | |
749 | write_seqcount_end(&inode->i_size_seqcount); | |
74e3d1e1 | 750 | preempt_enable(); |
1da177e4 LT |
751 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) |
752 | preempt_disable(); | |
753 | inode->i_size = i_size; | |
754 | preempt_enable(); | |
755 | #else | |
756 | inode->i_size = i_size; | |
757 | #endif | |
758 | } | |
759 | ||
92361636 EB |
760 | /* Helper functions so that in most cases filesystems will |
761 | * not need to deal directly with kuid_t and kgid_t and can | |
762 | * instead deal with the raw numeric values that are stored | |
763 | * in the filesystem. | |
764 | */ | |
765 | static inline uid_t i_uid_read(const struct inode *inode) | |
766 | { | |
767 | return from_kuid(&init_user_ns, inode->i_uid); | |
768 | } | |
769 | ||
770 | static inline gid_t i_gid_read(const struct inode *inode) | |
771 | { | |
772 | return from_kgid(&init_user_ns, inode->i_gid); | |
773 | } | |
774 | ||
775 | static inline void i_uid_write(struct inode *inode, uid_t uid) | |
776 | { | |
777 | inode->i_uid = make_kuid(&init_user_ns, uid); | |
778 | } | |
779 | ||
780 | static inline void i_gid_write(struct inode *inode, gid_t gid) | |
781 | { | |
782 | inode->i_gid = make_kgid(&init_user_ns, gid); | |
783 | } | |
784 | ||
48ed214d | 785 | static inline unsigned iminor(const struct inode *inode) |
1da177e4 LT |
786 | { |
787 | return MINOR(inode->i_rdev); | |
788 | } | |
789 | ||
48ed214d | 790 | static inline unsigned imajor(const struct inode *inode) |
1da177e4 LT |
791 | { |
792 | return MAJOR(inode->i_rdev); | |
793 | } | |
794 | ||
795 | extern struct block_device *I_BDEV(struct inode *inode); | |
796 | ||
797 | struct fown_struct { | |
798 | rwlock_t lock; /* protects pid, uid, euid fields */ | |
609d7fa9 EB |
799 | struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ |
800 | enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ | |
92361636 | 801 | kuid_t uid, euid; /* uid/euid of process setting the owner */ |
1da177e4 LT |
802 | int signum; /* posix.1b rt signal to be delivered on IO */ |
803 | }; | |
804 | ||
805 | /* | |
806 | * Track a single file's readahead state | |
807 | */ | |
808 | struct file_ra_state { | |
937085aa FW |
809 | pgoff_t start; /* where readahead started */ |
810 | unsigned int size; /* # of readahead pages */ | |
811 | unsigned int async_size; /* do asynchronous readahead when | |
f9acc8c7 | 812 | there are only # of pages ahead */ |
5ce1110b | 813 | |
937085aa | 814 | unsigned int ra_pages; /* Maximum readahead window */ |
1ebf26a9 | 815 | unsigned int mmap_miss; /* Cache miss stat for mmap accesses */ |
f4e6b498 | 816 | loff_t prev_pos; /* Cache last read() position */ |
1da177e4 | 817 | }; |
1da177e4 | 818 | |
5ce1110b FW |
819 | /* |
820 | * Check if @index falls in the readahead windows. | |
821 | */ | |
822 | static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) | |
823 | { | |
f9acc8c7 FW |
824 | return (index >= ra->start && |
825 | index < ra->start + ra->size); | |
5ce1110b FW |
826 | } |
827 | ||
1da177e4 | 828 | struct file { |
2f512016 | 829 | union { |
4f5e65a1 | 830 | struct llist_node fu_llist; |
2f512016 ED |
831 | struct rcu_head fu_rcuhead; |
832 | } f_u; | |
0f7fc9e4 | 833 | struct path f_path; |
dd37978c | 834 | struct inode *f_inode; /* cached value */ |
99ac48f5 | 835 | const struct file_operations *f_op; |
ef3d0fd2 AK |
836 | |
837 | /* | |
9c225f26 | 838 | * Protects f_ep_links, f_flags. |
ef3d0fd2 AK |
839 | * Must not be taken from IRQ context. |
840 | */ | |
841 | spinlock_t f_lock; | |
516e0cc5 | 842 | atomic_long_t f_count; |
1da177e4 | 843 | unsigned int f_flags; |
aeb5d727 | 844 | fmode_t f_mode; |
9c225f26 | 845 | struct mutex f_pos_lock; |
1da177e4 LT |
846 | loff_t f_pos; |
847 | struct fown_struct f_owner; | |
d76b0d9b | 848 | const struct cred *f_cred; |
1da177e4 LT |
849 | struct file_ra_state f_ra; |
850 | ||
2b47c361 | 851 | u64 f_version; |
50462062 | 852 | #ifdef CONFIG_SECURITY |
1da177e4 | 853 | void *f_security; |
50462062 | 854 | #endif |
1da177e4 LT |
855 | /* needed for tty driver, and maybe others */ |
856 | void *private_data; | |
857 | ||
858 | #ifdef CONFIG_EPOLL | |
859 | /* Used by fs/eventpoll.c to link all the hooks to this file */ | |
860 | struct list_head f_ep_links; | |
28d82dc1 | 861 | struct list_head f_tfile_llink; |
1da177e4 LT |
862 | #endif /* #ifdef CONFIG_EPOLL */ |
863 | struct address_space *f_mapping; | |
bd2a31d5 | 864 | } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ |
1da177e4 | 865 | |
990d6c2d AK |
866 | struct file_handle { |
867 | __u32 handle_bytes; | |
868 | int handle_type; | |
869 | /* file identifier */ | |
870 | unsigned char f_handle[0]; | |
871 | }; | |
872 | ||
cb0942b8 AV |
873 | static inline struct file *get_file(struct file *f) |
874 | { | |
875 | atomic_long_inc(&f->f_count); | |
876 | return f; | |
877 | } | |
90f31d0e | 878 | #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count) |
d7065da0 | 879 | #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1) |
516e0cc5 | 880 | #define file_count(x) atomic_long_read(&(x)->f_count) |
1da177e4 LT |
881 | |
882 | #define MAX_NON_LFS ((1UL<<31) - 1) | |
883 | ||
884 | /* Page cache limit. The filesystems should put that into their s_maxbytes | |
885 | limits, otherwise bad things can happen in VM. */ | |
886 | #if BITS_PER_LONG==32 | |
2bd2c194 | 887 | #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) |
1da177e4 | 888 | #elif BITS_PER_LONG==64 |
614c321f | 889 | #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL) |
1da177e4 LT |
890 | #endif |
891 | ||
892 | #define FL_POSIX 1 | |
893 | #define FL_FLOCK 2 | |
617588d5 | 894 | #define FL_DELEG 4 /* NFSv4 delegation */ |
1da177e4 | 895 | #define FL_ACCESS 8 /* not trying to lock, just looking */ |
f475ae95 | 896 | #define FL_EXISTS 16 /* when unlocking, test for existence */ |
1da177e4 | 897 | #define FL_LEASE 32 /* lease held on this file */ |
75e1fcc0 | 898 | #define FL_CLOSE 64 /* unlock on close */ |
1da177e4 | 899 | #define FL_SLEEP 128 /* A blocking lock */ |
778fc546 BF |
900 | #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */ |
901 | #define FL_UNLOCK_PENDING 512 /* Lease is being broken */ | |
cff2fce5 | 902 | #define FL_OFDLCK 1024 /* lock is "owned" by struct file */ |
11afe9f7 | 903 | #define FL_LAYOUT 2048 /* outstanding pNFS layout */ |
1da177e4 | 904 | |
bde74e4b MS |
905 | /* |
906 | * Special return value from posix_lock_file() and vfs_lock_file() for | |
907 | * asynchronous locking. | |
908 | */ | |
909 | #define FILE_LOCK_DEFERRED 1 | |
910 | ||
7ca76311 | 911 | /* legacy typedef, should eventually be removed */ |
17fa388d | 912 | typedef void *fl_owner_t; |
1da177e4 | 913 | |
a7231a97 JL |
914 | struct file_lock; |
915 | ||
1da177e4 | 916 | struct file_lock_operations { |
1da177e4 LT |
917 | void (*fl_copy_lock)(struct file_lock *, struct file_lock *); |
918 | void (*fl_release_private)(struct file_lock *); | |
919 | }; | |
920 | ||
921 | struct lock_manager_operations { | |
8fb47a4f | 922 | int (*lm_compare_owner)(struct file_lock *, struct file_lock *); |
3999e493 | 923 | unsigned long (*lm_owner_key)(struct file_lock *); |
cae80b30 JL |
924 | fl_owner_t (*lm_get_owner)(fl_owner_t); |
925 | void (*lm_put_owner)(fl_owner_t); | |
8fb47a4f | 926 | void (*lm_notify)(struct file_lock *); /* unblock callback */ |
d0449b90 | 927 | int (*lm_grant)(struct file_lock *, int); |
4d01b7f5 | 928 | bool (*lm_break)(struct file_lock *); |
7448cc37 | 929 | int (*lm_change)(struct file_lock *, int, struct list_head *); |
1c7dd2ff | 930 | void (*lm_setup)(struct file_lock *, void **); |
1da177e4 LT |
931 | }; |
932 | ||
af558e33 BF |
933 | struct lock_manager { |
934 | struct list_head list; | |
935 | }; | |
936 | ||
5ccb0066 SK |
937 | struct net; |
938 | void locks_start_grace(struct net *, struct lock_manager *); | |
af558e33 | 939 | void locks_end_grace(struct lock_manager *); |
5ccb0066 | 940 | int locks_in_grace(struct net *); |
af558e33 | 941 | |
1da177e4 LT |
942 | /* that will die - we need it for nfs_lock_info */ |
943 | #include <linux/nfs_fs_i.h> | |
944 | ||
1cb36012 JL |
945 | /* |
946 | * struct file_lock represents a generic "file lock". It's used to represent | |
947 | * POSIX byte range locks, BSD (flock) locks, and leases. It's important to | |
948 | * note that the same struct is used to represent both a request for a lock and | |
949 | * the lock itself, but the same object is never used for both. | |
950 | * | |
951 | * FIXME: should we create a separate "struct lock_request" to help distinguish | |
952 | * these two uses? | |
953 | * | |
8116bf4c | 954 | * The varous i_flctx lists are ordered by: |
1cb36012 | 955 | * |
8116bf4c JL |
956 | * 1) lock owner |
957 | * 2) lock range start | |
958 | * 3) lock range end | |
1cb36012 JL |
959 | * |
960 | * Obviously, the last two criteria only matter for POSIX locks. | |
961 | */ | |
1da177e4 LT |
962 | struct file_lock { |
963 | struct file_lock *fl_next; /* singly linked list for this inode */ | |
6dee60f6 | 964 | struct list_head fl_list; /* link into file_lock_context */ |
139ca04e | 965 | struct hlist_node fl_link; /* node in global lists */ |
1da177e4 LT |
966 | struct list_head fl_block; /* circular list of blocked processes */ |
967 | fl_owner_t fl_owner; | |
710b7216 | 968 | unsigned int fl_flags; |
afc1246f | 969 | unsigned char fl_type; |
1da177e4 | 970 | unsigned int fl_pid; |
7012b02a | 971 | int fl_link_cpu; /* what cpu's list is this on? */ |
ab1f1611 | 972 | struct pid *fl_nspid; |
1da177e4 LT |
973 | wait_queue_head_t fl_wait; |
974 | struct file *fl_file; | |
1da177e4 LT |
975 | loff_t fl_start; |
976 | loff_t fl_end; | |
977 | ||
978 | struct fasync_struct * fl_fasync; /* for lease break notifications */ | |
778fc546 BF |
979 | /* for lease breaks: */ |
980 | unsigned long fl_break_time; | |
981 | unsigned long fl_downgrade_time; | |
1da177e4 | 982 | |
6aed6285 | 983 | const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ |
7b021967 | 984 | const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ |
1da177e4 LT |
985 | union { |
986 | struct nfs_lock_info nfs_fl; | |
8d0a8a9d | 987 | struct nfs4_lock_info nfs4_fl; |
e8d6c554 DH |
988 | struct { |
989 | struct list_head link; /* link in AFS vnode's pending_locks list */ | |
990 | int state; /* state of grant or error if -ve */ | |
991 | } afs; | |
1da177e4 LT |
992 | } fl_u; |
993 | }; | |
994 | ||
4a075e39 | 995 | struct file_lock_context { |
6109c850 | 996 | spinlock_t flc_lock; |
4a075e39 | 997 | struct list_head flc_flock; |
bd61e0a9 | 998 | struct list_head flc_posix; |
8634b51f | 999 | struct list_head flc_lease; |
4a075e39 JL |
1000 | }; |
1001 | ||
1da177e4 LT |
1002 | /* The following constant reflects the upper bound of the file/locking space */ |
1003 | #ifndef OFFSET_MAX | |
1004 | #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) | |
1005 | #define OFFSET_MAX INT_LIMIT(loff_t) | |
1006 | #define OFFT_OFFSET_MAX INT_LIMIT(off_t) | |
1007 | #endif | |
1008 | ||
1da177e4 LT |
1009 | #include <linux/fcntl.h> |
1010 | ||
bfcd17a6 TP |
1011 | extern void send_sigio(struct fown_struct *fown, int fd, int band); |
1012 | ||
bfcd17a6 | 1013 | #ifdef CONFIG_FILE_LOCKING |
c1e62b8f | 1014 | extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *); |
c293621b PS |
1015 | extern int fcntl_setlk(unsigned int, struct file *, unsigned int, |
1016 | struct flock __user *); | |
1da177e4 LT |
1017 | |
1018 | #if BITS_PER_LONG == 32 | |
c1e62b8f | 1019 | extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *); |
c293621b PS |
1020 | extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, |
1021 | struct flock64 __user *); | |
1da177e4 LT |
1022 | #endif |
1023 | ||
1da177e4 LT |
1024 | extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); |
1025 | extern int fcntl_getlease(struct file *filp); | |
1026 | ||
1027 | /* fs/locks.c */ | |
4a075e39 | 1028 | void locks_free_lock_context(struct file_lock_context *ctx); |
05fa3135 | 1029 | void locks_free_lock(struct file_lock *fl); |
1da177e4 | 1030 | extern void locks_init_lock(struct file_lock *); |
c5b1f0d9 | 1031 | extern struct file_lock * locks_alloc_lock(void); |
1da177e4 | 1032 | extern void locks_copy_lock(struct file_lock *, struct file_lock *); |
3fe0fff1 | 1033 | extern void locks_copy_conflock(struct file_lock *, struct file_lock *); |
1da177e4 | 1034 | extern void locks_remove_posix(struct file *, fl_owner_t); |
78ed8a13 | 1035 | extern void locks_remove_file(struct file *); |
a9e61e25 | 1036 | extern void locks_release_private(struct file_lock *); |
6d34ac19 | 1037 | extern void posix_test_lock(struct file *, struct file_lock *); |
150b3934 | 1038 | extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); |
1da177e4 | 1039 | extern int posix_lock_file_wait(struct file *, struct file_lock *); |
f891a29f | 1040 | extern int posix_unblock_lock(struct file_lock *); |
3ee17abd | 1041 | extern int vfs_test_lock(struct file *, struct file_lock *); |
150b3934 | 1042 | extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); |
9b9d2ab4 | 1043 | extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); |
1da177e4 | 1044 | extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl); |
df4e8d2c | 1045 | extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type); |
1da177e4 | 1046 | extern void lease_get_mtime(struct inode *, struct timespec *time); |
e6f5c789 JL |
1047 | extern int generic_setlease(struct file *, long, struct file_lock **, void **priv); |
1048 | extern int vfs_setlease(struct file *, long, struct file_lock **, void **); | |
7448cc37 | 1049 | extern int lease_modify(struct file_lock *, int, struct list_head *); |
6c8c9031 AV |
1050 | struct files_struct; |
1051 | extern void show_fd_locks(struct seq_file *f, | |
1052 | struct file *filp, struct files_struct *files); | |
bfcd17a6 | 1053 | #else /* !CONFIG_FILE_LOCKING */ |
c1e62b8f JL |
1054 | static inline int fcntl_getlk(struct file *file, unsigned int cmd, |
1055 | struct flock __user *user) | |
c2aca5e5 SW |
1056 | { |
1057 | return -EINVAL; | |
1058 | } | |
1059 | ||
1060 | static inline int fcntl_setlk(unsigned int fd, struct file *file, | |
1061 | unsigned int cmd, struct flock __user *user) | |
1062 | { | |
1063 | return -EACCES; | |
1064 | } | |
1065 | ||
bfcd17a6 | 1066 | #if BITS_PER_LONG == 32 |
c1e62b8f JL |
1067 | static inline int fcntl_getlk64(struct file *file, unsigned int cmd, |
1068 | struct flock64 __user *user) | |
c2aca5e5 SW |
1069 | { |
1070 | return -EINVAL; | |
1071 | } | |
1072 | ||
1073 | static inline int fcntl_setlk64(unsigned int fd, struct file *file, | |
1074 | unsigned int cmd, struct flock64 __user *user) | |
1075 | { | |
1076 | return -EACCES; | |
1077 | } | |
bfcd17a6 | 1078 | #endif |
c2aca5e5 SW |
1079 | static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) |
1080 | { | |
1b2b32dc | 1081 | return -EINVAL; |
c2aca5e5 SW |
1082 | } |
1083 | ||
1084 | static inline int fcntl_getlease(struct file *filp) | |
1085 | { | |
1b2b32dc | 1086 | return F_UNLCK; |
c2aca5e5 SW |
1087 | } |
1088 | ||
4a075e39 JL |
1089 | static inline void |
1090 | locks_free_lock_context(struct file_lock_context *ctx) | |
1091 | { | |
1092 | } | |
1093 | ||
c2aca5e5 SW |
1094 | static inline void locks_init_lock(struct file_lock *fl) |
1095 | { | |
1096 | return; | |
1097 | } | |
1098 | ||
3fe0fff1 | 1099 | static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl) |
c2aca5e5 SW |
1100 | { |
1101 | return; | |
1102 | } | |
1103 | ||
1104 | static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) | |
1105 | { | |
1106 | return; | |
1107 | } | |
1108 | ||
1109 | static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) | |
1110 | { | |
1111 | return; | |
1112 | } | |
1113 | ||
78ed8a13 | 1114 | static inline void locks_remove_file(struct file *filp) |
c2aca5e5 SW |
1115 | { |
1116 | return; | |
1117 | } | |
1118 | ||
1119 | static inline void posix_test_lock(struct file *filp, struct file_lock *fl) | |
1120 | { | |
1121 | return; | |
1122 | } | |
1123 | ||
1124 | static inline int posix_lock_file(struct file *filp, struct file_lock *fl, | |
1125 | struct file_lock *conflock) | |
1126 | { | |
1127 | return -ENOLCK; | |
1128 | } | |
1129 | ||
1130 | static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl) | |
1131 | { | |
1132 | return -ENOLCK; | |
1133 | } | |
1134 | ||
f891a29f | 1135 | static inline int posix_unblock_lock(struct file_lock *waiter) |
c2aca5e5 SW |
1136 | { |
1137 | return -ENOENT; | |
1138 | } | |
1139 | ||
1140 | static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) | |
1141 | { | |
1142 | return 0; | |
1143 | } | |
1144 | ||
1145 | static inline int vfs_lock_file(struct file *filp, unsigned int cmd, | |
1146 | struct file_lock *fl, struct file_lock *conf) | |
1147 | { | |
1148 | return -ENOLCK; | |
1149 | } | |
1150 | ||
1151 | static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) | |
1152 | { | |
1153 | return 0; | |
1154 | } | |
1155 | ||
1156 | static inline int flock_lock_file_wait(struct file *filp, | |
1157 | struct file_lock *request) | |
1158 | { | |
1159 | return -ENOLCK; | |
1160 | } | |
1161 | ||
df4e8d2c | 1162 | static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type) |
c2aca5e5 SW |
1163 | { |
1164 | return 0; | |
1165 | } | |
1166 | ||
1167 | static inline void lease_get_mtime(struct inode *inode, struct timespec *time) | |
1168 | { | |
1169 | return; | |
1170 | } | |
1171 | ||
1172 | static inline int generic_setlease(struct file *filp, long arg, | |
e6f5c789 | 1173 | struct file_lock **flp, void **priv) |
c2aca5e5 SW |
1174 | { |
1175 | return -EINVAL; | |
1176 | } | |
1177 | ||
1178 | static inline int vfs_setlease(struct file *filp, long arg, | |
e6f5c789 | 1179 | struct file_lock **lease, void **priv) |
c2aca5e5 SW |
1180 | { |
1181 | return -EINVAL; | |
1182 | } | |
1183 | ||
7448cc37 | 1184 | static inline int lease_modify(struct file_lock *fl, int arg, |
c45198ed | 1185 | struct list_head *dispose) |
c2aca5e5 SW |
1186 | { |
1187 | return -EINVAL; | |
1188 | } | |
6c8c9031 AV |
1189 | |
1190 | struct files_struct; | |
1191 | static inline void show_fd_locks(struct seq_file *f, | |
1192 | struct file *filp, struct files_struct *files) {} | |
bfcd17a6 TP |
1193 | #endif /* !CONFIG_FILE_LOCKING */ |
1194 | ||
1da177e4 LT |
1195 | |
1196 | struct fasync_struct { | |
989a2979 ED |
1197 | spinlock_t fa_lock; |
1198 | int magic; | |
1199 | int fa_fd; | |
1200 | struct fasync_struct *fa_next; /* singly linked list */ | |
1201 | struct file *fa_file; | |
1202 | struct rcu_head fa_rcu; | |
1da177e4 LT |
1203 | }; |
1204 | ||
1205 | #define FASYNC_MAGIC 0x4601 | |
1206 | ||
1207 | /* SMP safe fasync helpers: */ | |
1208 | extern int fasync_helper(int, struct file *, int, struct fasync_struct **); | |
f7347ce4 LT |
1209 | extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); |
1210 | extern int fasync_remove_entry(struct file *, struct fasync_struct **); | |
1211 | extern struct fasync_struct *fasync_alloc(void); | |
1212 | extern void fasync_free(struct fasync_struct *); | |
1213 | ||
1da177e4 LT |
1214 | /* can be called from interrupts */ |
1215 | extern void kill_fasync(struct fasync_struct **, int, int); | |
1da177e4 | 1216 | |
e0b93edd JL |
1217 | extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); |
1218 | extern void f_setown(struct file *filp, unsigned long arg, int force); | |
1da177e4 | 1219 | extern void f_delown(struct file *filp); |
609d7fa9 | 1220 | extern pid_t f_getown(struct file *filp); |
1da177e4 LT |
1221 | extern int send_sigurg(struct fown_struct *fown); |
1222 | ||
5accdf82 JK |
1223 | struct mm_struct; |
1224 | ||
1da177e4 LT |
1225 | /* |
1226 | * Umount options | |
1227 | */ | |
1228 | ||
1229 | #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ | |
1230 | #define MNT_DETACH 0x00000002 /* Just detach from the tree */ | |
1231 | #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ | |
db1f05bb MS |
1232 | #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ |
1233 | #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ | |
1da177e4 | 1234 | |
1da177e4 | 1235 | |
5accdf82 JK |
1236 | /* Possible states of 'frozen' field */ |
1237 | enum { | |
1238 | SB_UNFROZEN = 0, /* FS is unfrozen */ | |
1239 | SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ | |
5accdf82 JK |
1240 | SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ |
1241 | SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop | |
1242 | * internal threads if needed) */ | |
1243 | SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ | |
1244 | }; | |
1245 | ||
1246 | #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) | |
1247 | ||
1248 | struct sb_writers { | |
1249 | /* Counters for counting writers at each level */ | |
1250 | struct percpu_counter counter[SB_FREEZE_LEVELS]; | |
1251 | wait_queue_head_t wait; /* queue for waiting for | |
1252 | writers / faults to finish */ | |
1253 | int frozen; /* Is sb frozen? */ | |
1254 | wait_queue_head_t wait_unfrozen; /* queue for waiting for | |
1255 | sb to be thawed */ | |
1256 | #ifdef CONFIG_DEBUG_LOCK_ALLOC | |
1257 | struct lockdep_map lock_map[SB_FREEZE_LEVELS]; | |
1258 | #endif | |
1259 | }; | |
1260 | ||
1da177e4 LT |
1261 | struct super_block { |
1262 | struct list_head s_list; /* Keep this first */ | |
1263 | dev_t s_dev; /* search index; _not_ kdev_t */ | |
270ba5f7 RK |
1264 | unsigned char s_blocksize_bits; |
1265 | unsigned long s_blocksize; | |
42cb56ae | 1266 | loff_t s_maxbytes; /* Max file size */ |
1da177e4 | 1267 | struct file_system_type *s_type; |
ee9b6d61 | 1268 | const struct super_operations *s_op; |
61e225dc | 1269 | const struct dquot_operations *dq_op; |
0d54b217 | 1270 | const struct quotactl_ops *s_qcop; |
39655164 | 1271 | const struct export_operations *s_export_op; |
1da177e4 LT |
1272 | unsigned long s_flags; |
1273 | unsigned long s_magic; | |
1274 | struct dentry *s_root; | |
1275 | struct rw_semaphore s_umount; | |
1da177e4 | 1276 | int s_count; |
1da177e4 | 1277 | atomic_t s_active; |
50462062 | 1278 | #ifdef CONFIG_SECURITY |
1da177e4 | 1279 | void *s_security; |
50462062 | 1280 | #endif |
bb435453 | 1281 | const struct xattr_handler **s_xattr; |
1da177e4 LT |
1282 | |
1283 | struct list_head s_inodes; /* all inodes */ | |
ceb5bdc2 | 1284 | struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */ |
39f7c4db | 1285 | struct list_head s_mounts; /* list of mounts; _not_ for fs use */ |
1da177e4 | 1286 | struct block_device *s_bdev; |
32a88aa1 | 1287 | struct backing_dev_info *s_bdi; |
acaebfd8 | 1288 | struct mtd_info *s_mtd; |
a5166169 | 1289 | struct hlist_node s_instances; |
2c5f648a | 1290 | unsigned int s_quota_types; /* Bitmask of supported quota types */ |
1da177e4 LT |
1291 | struct quota_info s_dquot; /* Diskquota specific options */ |
1292 | ||
5accdf82 | 1293 | struct sb_writers s_writers; |
1da177e4 LT |
1294 | |
1295 | char s_id[32]; /* Informational name */ | |
0bba0169 | 1296 | u8 s_uuid[16]; /* UUID */ |
1da177e4 LT |
1297 | |
1298 | void *s_fs_info; /* Filesystem private info */ | |
8de52778 | 1299 | unsigned int s_max_links; |
30c40d2c | 1300 | fmode_t s_mode; |
1da177e4 | 1301 | |
270ba5f7 RK |
1302 | /* Granularity of c/m/atime in ns. |
1303 | Cannot be worse than a second */ | |
1304 | u32 s_time_gran; | |
1305 | ||
1da177e4 LT |
1306 | /* |
1307 | * The next field is for VFS *only*. No filesystems have any business | |
1308 | * even looking at it. You had been warned. | |
1309 | */ | |
a11f3a05 | 1310 | struct mutex s_vfs_rename_mutex; /* Kludge */ |
1da177e4 | 1311 | |
79c0b2df MS |
1312 | /* |
1313 | * Filesystem subtype. If non-empty the filesystem type field | |
1314 | * in /proc/mounts will be "type.subtype" | |
1315 | */ | |
1316 | char *s_subtype; | |
b3b304a2 MS |
1317 | |
1318 | /* | |
1319 | * Saved mount options for lazy filesystems using | |
1320 | * generic_show_options() | |
1321 | */ | |
4d2deb40 | 1322 | char __rcu *s_options; |
c8aebb0c | 1323 | const struct dentry_operations *s_d_op; /* default d_op for dentries */ |
9fdfdcf1 DM |
1324 | |
1325 | /* | |
1326 | * Saved pool identifier for cleancache (-1 means none) | |
1327 | */ | |
1328 | int cleancache_poolid; | |
b0d40c92 DC |
1329 | |
1330 | struct shrinker s_shrink; /* per-sb shrinker handle */ | |
4ed5e82f | 1331 | |
7ada4db8 MS |
1332 | /* Number of inodes with nlink == 0 but still referenced */ |
1333 | atomic_long_t s_remove_count; | |
1334 | ||
4ed5e82f MS |
1335 | /* Being remounted read-only */ |
1336 | int s_readonly_remount; | |
7b7a8665 CH |
1337 | |
1338 | /* AIO completions deferred from interrupt context */ | |
1339 | struct workqueue_struct *s_dio_done_wq; | |
215752fc | 1340 | struct hlist_head s_pins; |
f6041567 DC |
1341 | |
1342 | /* | |
1343 | * Keep the lru lists last in the structure so they always sit on their | |
1344 | * own individual cachelines. | |
1345 | */ | |
1346 | struct list_lru s_dentry_lru ____cacheline_aligned_in_smp; | |
1347 | struct list_lru s_inode_lru ____cacheline_aligned_in_smp; | |
e2fec7c3 | 1348 | struct rcu_head rcu; |
69c433ed MS |
1349 | |
1350 | /* | |
1351 | * Indicates how deep in a filesystem stack this SB is | |
1352 | */ | |
1353 | int s_stack_depth; | |
1da177e4 LT |
1354 | }; |
1355 | ||
1356 | extern struct timespec current_fs_time(struct super_block *sb); | |
1357 | ||
1358 | /* | |
1359 | * Snapshotting support. | |
1360 | */ | |
1da177e4 | 1361 | |
5accdf82 JK |
1362 | void __sb_end_write(struct super_block *sb, int level); |
1363 | int __sb_start_write(struct super_block *sb, int level, bool wait); | |
1364 | ||
1365 | /** | |
1366 | * sb_end_write - drop write access to a superblock | |
1367 | * @sb: the super we wrote to | |
1368 | * | |
1369 | * Decrement number of writers to the filesystem. Wake up possible waiters | |
1370 | * wanting to freeze the filesystem. | |
1371 | */ | |
1372 | static inline void sb_end_write(struct super_block *sb) | |
1373 | { | |
1374 | __sb_end_write(sb, SB_FREEZE_WRITE); | |
1375 | } | |
1376 | ||
1377 | /** | |
1378 | * sb_end_pagefault - drop write access to a superblock from a page fault | |
1379 | * @sb: the super we wrote to | |
1380 | * | |
1381 | * Decrement number of processes handling write page fault to the filesystem. | |
1382 | * Wake up possible waiters wanting to freeze the filesystem. | |
1383 | */ | |
1384 | static inline void sb_end_pagefault(struct super_block *sb) | |
1385 | { | |
1386 | __sb_end_write(sb, SB_FREEZE_PAGEFAULT); | |
1387 | } | |
1388 | ||
1389 | /** | |
1390 | * sb_end_intwrite - drop write access to a superblock for internal fs purposes | |
1391 | * @sb: the super we wrote to | |
1392 | * | |
1393 | * Decrement fs-internal number of writers to the filesystem. Wake up possible | |
1394 | * waiters wanting to freeze the filesystem. | |
1395 | */ | |
1396 | static inline void sb_end_intwrite(struct super_block *sb) | |
1397 | { | |
1398 | __sb_end_write(sb, SB_FREEZE_FS); | |
1399 | } | |
1400 | ||
1401 | /** | |
1402 | * sb_start_write - get write access to a superblock | |
1403 | * @sb: the super we write to | |
1404 | * | |
1405 | * When a process wants to write data or metadata to a file system (i.e. dirty | |
1406 | * a page or an inode), it should embed the operation in a sb_start_write() - | |
1407 | * sb_end_write() pair to get exclusion against file system freezing. This | |
1408 | * function increments number of writers preventing freezing. If the file | |
1409 | * system is already frozen, the function waits until the file system is | |
1410 | * thawed. | |
1411 | * | |
1412 | * Since freeze protection behaves as a lock, users have to preserve | |
1413 | * ordering of freeze protection and other filesystem locks. Generally, | |
1414 | * freeze protection should be the outermost lock. In particular, we have: | |
1415 | * | |
1416 | * sb_start_write | |
1417 | * -> i_mutex (write path, truncate, directory ops, ...) | |
1418 | * -> s_umount (freeze_super, thaw_super) | |
1419 | */ | |
1420 | static inline void sb_start_write(struct super_block *sb) | |
1421 | { | |
1422 | __sb_start_write(sb, SB_FREEZE_WRITE, true); | |
1423 | } | |
1424 | ||
1425 | static inline int sb_start_write_trylock(struct super_block *sb) | |
1426 | { | |
1427 | return __sb_start_write(sb, SB_FREEZE_WRITE, false); | |
1428 | } | |
1429 | ||
1430 | /** | |
1431 | * sb_start_pagefault - get write access to a superblock from a page fault | |
1432 | * @sb: the super we write to | |
1433 | * | |
1434 | * When a process starts handling write page fault, it should embed the | |
1435 | * operation into sb_start_pagefault() - sb_end_pagefault() pair to get | |
1436 | * exclusion against file system freezing. This is needed since the page fault | |
1437 | * is going to dirty a page. This function increments number of running page | |
1438 | * faults preventing freezing. If the file system is already frozen, the | |
1439 | * function waits until the file system is thawed. | |
1440 | * | |
1441 | * Since page fault freeze protection behaves as a lock, users have to preserve | |
1442 | * ordering of freeze protection and other filesystem locks. It is advised to | |
1443 | * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault | |
1444 | * handling code implies lock dependency: | |
1445 | * | |
1446 | * mmap_sem | |
1447 | * -> sb_start_pagefault | |
1448 | */ | |
1449 | static inline void sb_start_pagefault(struct super_block *sb) | |
1450 | { | |
1451 | __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true); | |
1452 | } | |
1453 | ||
1454 | /* | |
1455 | * sb_start_intwrite - get write access to a superblock for internal fs purposes | |
1456 | * @sb: the super we write to | |
1457 | * | |
1458 | * This is the third level of protection against filesystem freezing. It is | |
1459 | * free for use by a filesystem. The only requirement is that it must rank | |
1460 | * below sb_start_pagefault. | |
1461 | * | |
1462 | * For example filesystem can call sb_start_intwrite() when starting a | |
1463 | * transaction which somewhat eases handling of freezing for internal sources | |
1464 | * of filesystem changes (internal fs threads, discarding preallocation on file | |
1465 | * close, etc.). | |
1466 | */ | |
1467 | static inline void sb_start_intwrite(struct super_block *sb) | |
1468 | { | |
1469 | __sb_start_write(sb, SB_FREEZE_FS, true); | |
1470 | } | |
1da177e4 | 1471 | |
1da177e4 | 1472 | |
2e149670 | 1473 | extern bool inode_owner_or_capable(const struct inode *inode); |
3bd858ab | 1474 | |
1da177e4 LT |
1475 | /* |
1476 | * VFS helper functions.. | |
1477 | */ | |
312b63fb | 1478 | extern int vfs_create(struct inode *, struct dentry *, umode_t, bool); |
18bb1db3 | 1479 | extern int vfs_mkdir(struct inode *, struct dentry *, umode_t); |
1a67aafb | 1480 | extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t); |
db2e747b | 1481 | extern int vfs_symlink(struct inode *, struct dentry *, const char *); |
146a8595 | 1482 | extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **); |
1da177e4 | 1483 | extern int vfs_rmdir(struct inode *, struct dentry *); |
b21996e3 | 1484 | extern int vfs_unlink(struct inode *, struct dentry *, struct inode **); |
520c8b16 | 1485 | extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int); |
787fb6bc | 1486 | extern int vfs_whiteout(struct inode *, struct dentry *); |
1da177e4 LT |
1487 | |
1488 | /* | |
1489 | * VFS dentry helper functions. | |
1490 | */ | |
1491 | extern void dentry_unhash(struct dentry *dentry); | |
1492 | ||
8c744fb8 CH |
1493 | /* |
1494 | * VFS file helper functions. | |
1495 | */ | |
a1bd120d | 1496 | extern void inode_init_owner(struct inode *inode, const struct inode *dir, |
62bb1091 | 1497 | umode_t mode); |
c4b929b8 MF |
1498 | /* |
1499 | * VFS FS_IOC_FIEMAP helper definitions. | |
1500 | */ | |
1501 | struct fiemap_extent_info { | |
1502 | unsigned int fi_flags; /* Flags as passed from user */ | |
1503 | unsigned int fi_extents_mapped; /* Number of mapped extents */ | |
1504 | unsigned int fi_extents_max; /* Size of fiemap_extent array */ | |
ecf5632d NK |
1505 | struct fiemap_extent __user *fi_extents_start; /* Start of |
1506 | fiemap_extent array */ | |
c4b929b8 MF |
1507 | }; |
1508 | int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical, | |
1509 | u64 phys, u64 len, u32 flags); | |
1510 | int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags); | |
1511 | ||
1da177e4 LT |
1512 | /* |
1513 | * File types | |
1514 | * | |
1515 | * NOTE! These match bits 12..15 of stat.st_mode | |
1516 | * (ie "(i_mode >> 12) & 15"). | |
1517 | */ | |
1518 | #define DT_UNKNOWN 0 | |
1519 | #define DT_FIFO 1 | |
1520 | #define DT_CHR 2 | |
1521 | #define DT_DIR 4 | |
1522 | #define DT_BLK 6 | |
1523 | #define DT_REG 8 | |
1524 | #define DT_LNK 10 | |
1525 | #define DT_SOCK 12 | |
1526 | #define DT_WHT 14 | |
1527 | ||
1da177e4 LT |
1528 | /* |
1529 | * This is the "filldir" function type, used by readdir() to let | |
1530 | * the kernel specify what kind of dirent layout it wants to have. | |
1531 | * This allows the kernel to read directories into kernel space or | |
1532 | * to have different dirent layouts depending on the binary type. | |
1533 | */ | |
ac7576f4 MS |
1534 | struct dir_context; |
1535 | typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, | |
1536 | unsigned); | |
1537 | ||
5c0ba4e0 | 1538 | struct dir_context { |
ac6614b7 | 1539 | const filldir_t actor; |
bb6f619b | 1540 | loff_t pos; |
5c0ba4e0 | 1541 | }; |
bb6f619b | 1542 | |
08f85851 | 1543 | struct block_device_operations; |
1da177e4 | 1544 | |
1da177e4 LT |
1545 | /* These macros are for out of kernel modules to test that |
1546 | * the kernel supports the unlocked_ioctl and compat_ioctl | |
1547 | * fields in struct file_operations. */ | |
1548 | #define HAVE_COMPAT_IOCTL 1 | |
1549 | #define HAVE_UNLOCKED_IOCTL 1 | |
1550 | ||
b4caecd4 CH |
1551 | /* |
1552 | * These flags let !MMU mmap() govern direct device mapping vs immediate | |
1553 | * copying more easily for MAP_PRIVATE, especially for ROM filesystems. | |
1554 | * | |
1555 | * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) | |
1556 | * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) | |
1557 | * NOMMU_MAP_READ: Can be mapped for reading | |
1558 | * NOMMU_MAP_WRITE: Can be mapped for writing | |
1559 | * NOMMU_MAP_EXEC: Can be mapped for execution | |
1560 | */ | |
1561 | #define NOMMU_MAP_COPY 0x00000001 | |
1562 | #define NOMMU_MAP_DIRECT 0x00000008 | |
1563 | #define NOMMU_MAP_READ VM_MAYREAD | |
1564 | #define NOMMU_MAP_WRITE VM_MAYWRITE | |
1565 | #define NOMMU_MAP_EXEC VM_MAYEXEC | |
1566 | ||
1567 | #define NOMMU_VMFLAGS \ | |
1568 | (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) | |
1569 | ||
1570 | ||
293bc982 AV |
1571 | struct iov_iter; |
1572 | ||
1da177e4 LT |
1573 | struct file_operations { |
1574 | struct module *owner; | |
1575 | loff_t (*llseek) (struct file *, loff_t, int); | |
1576 | ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); | |
1da177e4 | 1577 | ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); |
293bc982 AV |
1578 | ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); |
1579 | ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); | |
bb6f619b | 1580 | int (*iterate) (struct file *, struct dir_context *); |
1da177e4 | 1581 | unsigned int (*poll) (struct file *, struct poll_table_struct *); |
1da177e4 LT |
1582 | long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); |
1583 | long (*compat_ioctl) (struct file *, unsigned int, unsigned long); | |
1584 | int (*mmap) (struct file *, struct vm_area_struct *); | |
b2edffdd | 1585 | int (*mremap)(struct file *, struct vm_area_struct *); |
1da177e4 | 1586 | int (*open) (struct inode *, struct file *); |
75e1fcc0 | 1587 | int (*flush) (struct file *, fl_owner_t id); |
1da177e4 | 1588 | int (*release) (struct inode *, struct file *); |
02c24a82 | 1589 | int (*fsync) (struct file *, loff_t, loff_t, int datasync); |
1da177e4 LT |
1590 | int (*aio_fsync) (struct kiocb *, int datasync); |
1591 | int (*fasync) (int, struct file *, int); | |
1592 | int (*lock) (struct file *, int, struct file_lock *); | |
1da177e4 LT |
1593 | ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); |
1594 | unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1595 | int (*check_flags)(int); | |
1da177e4 | 1596 | int (*flock) (struct file *, int, struct file_lock *); |
cbb7e577 JA |
1597 | ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); |
1598 | ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); | |
e6f5c789 | 1599 | int (*setlease)(struct file *, long, struct file_lock **, void **); |
2fe17c10 CH |
1600 | long (*fallocate)(struct file *file, int mode, loff_t offset, |
1601 | loff_t len); | |
a3816ab0 | 1602 | void (*show_fdinfo)(struct seq_file *m, struct file *f); |
b4caecd4 CH |
1603 | #ifndef CONFIG_MMU |
1604 | unsigned (*mmap_capabilities)(struct file *); | |
1605 | #endif | |
1da177e4 LT |
1606 | }; |
1607 | ||
1608 | struct inode_operations { | |
00cd8dd3 | 1609 | struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); |
6e77137b | 1610 | const char * (*follow_link) (struct dentry *, void **); |
10556cb2 | 1611 | int (*permission) (struct inode *, int); |
4e34e719 | 1612 | struct posix_acl * (*get_acl)(struct inode *, int); |
44a7d7a8 NP |
1613 | |
1614 | int (*readlink) (struct dentry *, char __user *,int); | |
5f2c4179 | 1615 | void (*put_link) (struct inode *, void *); |
44a7d7a8 | 1616 | |
ebfc3b49 | 1617 | int (*create) (struct inode *,struct dentry *, umode_t, bool); |
1da177e4 LT |
1618 | int (*link) (struct dentry *,struct inode *,struct dentry *); |
1619 | int (*unlink) (struct inode *,struct dentry *); | |
1620 | int (*symlink) (struct inode *,struct dentry *,const char *); | |
18bb1db3 | 1621 | int (*mkdir) (struct inode *,struct dentry *,umode_t); |
1da177e4 | 1622 | int (*rmdir) (struct inode *,struct dentry *); |
1a67aafb | 1623 | int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t); |
1da177e4 LT |
1624 | int (*rename) (struct inode *, struct dentry *, |
1625 | struct inode *, struct dentry *); | |
520c8b16 MS |
1626 | int (*rename2) (struct inode *, struct dentry *, |
1627 | struct inode *, struct dentry *, unsigned int); | |
1da177e4 LT |
1628 | int (*setattr) (struct dentry *, struct iattr *); |
1629 | int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *); | |
1630 | int (*setxattr) (struct dentry *, const char *,const void *,size_t,int); | |
1631 | ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t); | |
1632 | ssize_t (*listxattr) (struct dentry *, char *, size_t); | |
1633 | int (*removexattr) (struct dentry *, const char *); | |
c4b929b8 MF |
1634 | int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, |
1635 | u64 len); | |
c3b2da31 | 1636 | int (*update_time)(struct inode *, struct timespec *, int); |
d9585277 | 1637 | int (*atomic_open)(struct inode *, struct dentry *, |
30d90494 | 1638 | struct file *, unsigned open_flag, |
d9585277 | 1639 | umode_t create_mode, int *opened); |
60545d0d | 1640 | int (*tmpfile) (struct inode *, struct dentry *, umode_t); |
893d46e4 | 1641 | int (*set_acl)(struct inode *, struct posix_acl *, int); |
4aa7c634 MS |
1642 | |
1643 | /* WARNING: probably going away soon, do not use! */ | |
44a7d7a8 | 1644 | } ____cacheline_aligned; |
1da177e4 | 1645 | |
eed4e51f | 1646 | ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, |
fcf63409 CY |
1647 | unsigned long nr_segs, unsigned long fast_segs, |
1648 | struct iovec *fast_pointer, | |
ac34ebb3 | 1649 | struct iovec **ret_pointer); |
eed4e51f | 1650 | |
6fb5032e | 1651 | extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *); |
493c84c0 | 1652 | extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *); |
1da177e4 LT |
1653 | extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); |
1654 | extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); | |
1655 | extern ssize_t vfs_readv(struct file *, const struct iovec __user *, | |
1656 | unsigned long, loff_t *); | |
1657 | extern ssize_t vfs_writev(struct file *, const struct iovec __user *, | |
1658 | unsigned long, loff_t *); | |
1659 | ||
1da177e4 LT |
1660 | struct super_operations { |
1661 | struct inode *(*alloc_inode)(struct super_block *sb); | |
1662 | void (*destroy_inode)(struct inode *); | |
1663 | ||
aa385729 | 1664 | void (*dirty_inode) (struct inode *, int flags); |
a9185b41 | 1665 | int (*write_inode) (struct inode *, struct writeback_control *wbc); |
45321ac5 | 1666 | int (*drop_inode) (struct inode *); |
be7ce416 | 1667 | void (*evict_inode) (struct inode *); |
1da177e4 | 1668 | void (*put_super) (struct super_block *); |
1da177e4 | 1669 | int (*sync_fs)(struct super_block *sb, int wait); |
48b6bca6 | 1670 | int (*freeze_super) (struct super_block *); |
c4be0c1d | 1671 | int (*freeze_fs) (struct super_block *); |
48b6bca6 | 1672 | int (*thaw_super) (struct super_block *); |
c4be0c1d | 1673 | int (*unfreeze_fs) (struct super_block *); |
726c3342 | 1674 | int (*statfs) (struct dentry *, struct kstatfs *); |
1da177e4 | 1675 | int (*remount_fs) (struct super_block *, int *, char *); |
42faad99 | 1676 | void (*umount_begin) (struct super_block *); |
1da177e4 | 1677 | |
34c80b1d | 1678 | int (*show_options)(struct seq_file *, struct dentry *); |
d861c630 | 1679 | int (*show_devname)(struct seq_file *, struct dentry *); |
a6322de6 | 1680 | int (*show_path)(struct seq_file *, struct dentry *); |
64132379 | 1681 | int (*show_stats)(struct seq_file *, struct dentry *); |
0e51a720 | 1682 | #ifdef CONFIG_QUOTA |
1da177e4 LT |
1683 | ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); |
1684 | ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); | |
2d0fa467 | 1685 | struct dquot **(*get_dquots)(struct inode *); |
0e51a720 | 1686 | #endif |
87d8fe1e | 1687 | int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t); |
4101b624 VD |
1688 | long (*nr_cached_objects)(struct super_block *, |
1689 | struct shrink_control *); | |
1690 | long (*free_cached_objects)(struct super_block *, | |
1691 | struct shrink_control *); | |
1da177e4 LT |
1692 | }; |
1693 | ||
bbc1096a DH |
1694 | /* |
1695 | * Inode flags - they have no relation to superblock flags now | |
1696 | */ | |
1697 | #define S_SYNC 1 /* Writes are synced at once */ | |
1698 | #define S_NOATIME 2 /* Do not update access times */ | |
1699 | #define S_APPEND 4 /* Append-only file */ | |
1700 | #define S_IMMUTABLE 8 /* Immutable file */ | |
1701 | #define S_DEAD 16 /* removed, but still open directory */ | |
1702 | #define S_NOQUOTA 32 /* Inode is not counted to quota */ | |
1703 | #define S_DIRSYNC 64 /* Directory modifications are synchronous */ | |
1704 | #define S_NOCMTIME 128 /* Do not update file c/mtime */ | |
1705 | #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ | |
1706 | #define S_PRIVATE 512 /* Inode is fs-internal */ | |
1707 | #define S_IMA 1024 /* Inode has an associated IMA struct */ | |
1708 | #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */ | |
1709 | #define S_NOSEC 4096 /* no suid or xattr security attributes */ | |
6cd176a5 | 1710 | #ifdef CONFIG_FS_DAX |
fbbbad4b MW |
1711 | #define S_DAX 8192 /* Direct Access, avoiding the page cache */ |
1712 | #else | |
1713 | #define S_DAX 0 /* Make all the DAX code disappear */ | |
1714 | #endif | |
bbc1096a DH |
1715 | |
1716 | /* | |
1717 | * Note that nosuid etc flags are inode-specific: setting some file-system | |
1718 | * flags just means all the inodes inherit those flags by default. It might be | |
1719 | * possible to override it selectively if you really wanted to with some | |
1720 | * ioctl() that is not currently implemented. | |
1721 | * | |
1722 | * Exception: MS_RDONLY is always applied to the entire file system. | |
1723 | * | |
1724 | * Unfortunately, it is possible to change a filesystems flags with it mounted | |
1725 | * with files in use. This means that all of the inodes will not have their | |
1726 | * i_flags updated. Hence, i_flags no longer inherit the superblock mount | |
1727 | * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org | |
1728 | */ | |
1729 | #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) | |
1730 | ||
1731 | #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY) | |
1732 | #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \ | |
1733 | ((inode)->i_flags & S_SYNC)) | |
1734 | #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \ | |
1735 | ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) | |
1736 | #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK) | |
1737 | #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME) | |
1738 | #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION) | |
1739 | ||
1740 | #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) | |
1741 | #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) | |
1742 | #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) | |
1743 | #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL) | |
1744 | ||
1745 | #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) | |
1746 | #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) | |
1747 | #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) | |
1748 | #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) | |
1749 | #define IS_IMA(inode) ((inode)->i_flags & S_IMA) | |
1750 | #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) | |
1751 | #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) | |
fbbbad4b | 1752 | #define IS_DAX(inode) ((inode)->i_flags & S_DAX) |
bbc1096a | 1753 | |
787fb6bc MS |
1754 | #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ |
1755 | (inode)->i_rdev == WHITEOUT_DEV) | |
1756 | ||
1c0eeaf5 | 1757 | /* |
250df6ed | 1758 | * Inode state bits. Protected by inode->i_lock |
1c0eeaf5 JE |
1759 | * |
1760 | * Three bits determine the dirty state of the inode, I_DIRTY_SYNC, | |
1761 | * I_DIRTY_DATASYNC and I_DIRTY_PAGES. | |
1762 | * | |
1763 | * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, | |
1764 | * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at | |
1765 | * various stages of removing an inode. | |
1766 | * | |
eaff8079 | 1767 | * Two bits are used for locking and completion notification, I_NEW and I_SYNC. |
1c0eeaf5 | 1768 | * |
e7ca2d41 JE |
1769 | * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on |
1770 | * fdatasync(). i_atime is the usual cause. | |
000cb48e JK |
1771 | * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of |
1772 | * these changes separately from I_DIRTY_SYNC so that we | |
1773 | * don't have to write inode on fdatasync() when only | |
1774 | * mtime has changed in it. | |
1c0eeaf5 | 1775 | * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. |
eaff8079 CH |
1776 | * I_NEW Serves as both a mutex and completion notification. |
1777 | * New inodes set I_NEW. If two processes both create | |
1778 | * the same inode, one of them will release its inode and | |
1779 | * wait for I_NEW to be released before returning. | |
1780 | * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can | |
1781 | * also cause waiting on I_NEW, without I_NEW actually | |
1782 | * being set. find_inode() uses this to prevent returning | |
1783 | * nearly-dead inodes. | |
1c0eeaf5 JE |
1784 | * I_WILL_FREE Must be set when calling write_inode_now() if i_count |
1785 | * is zero. I_FREEING must be set when I_WILL_FREE is | |
1786 | * cleared. | |
1787 | * I_FREEING Set when inode is about to be freed but still has dirty | |
1788 | * pages or buffers attached or the inode itself is still | |
1789 | * dirty. | |
dbd5768f JK |
1790 | * I_CLEAR Added by clear_inode(). In this state the inode is |
1791 | * clean and can be destroyed. Inode keeps I_FREEING. | |
1c0eeaf5 JE |
1792 | * |
1793 | * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are | |
1794 | * prohibited for many purposes. iget() must wait for | |
1795 | * the inode to be completely released, then create it | |
1796 | * anew. Other functions will just ignore such inodes, | |
eaff8079 | 1797 | * if appropriate. I_NEW is used for waiting. |
1c0eeaf5 | 1798 | * |
169ebd90 JK |
1799 | * I_SYNC Writeback of inode is running. The bit is set during |
1800 | * data writeback, and cleared with a wakeup on the bit | |
1801 | * address once it is done. The bit is also used to pin | |
1802 | * the inode in memory for flusher thread. | |
1c0eeaf5 | 1803 | * |
bd5fe6c5 CH |
1804 | * I_REFERENCED Marks the inode as recently references on the LRU list. |
1805 | * | |
1806 | * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). | |
1807 | * | |
1c0eeaf5 | 1808 | * Q: What is the difference between I_WILL_FREE and I_FREEING? |
1c0eeaf5 | 1809 | */ |
9e38d86f NP |
1810 | #define I_DIRTY_SYNC (1 << 0) |
1811 | #define I_DIRTY_DATASYNC (1 << 1) | |
1812 | #define I_DIRTY_PAGES (1 << 2) | |
eaff8079 CH |
1813 | #define __I_NEW 3 |
1814 | #define I_NEW (1 << __I_NEW) | |
9e38d86f NP |
1815 | #define I_WILL_FREE (1 << 4) |
1816 | #define I_FREEING (1 << 5) | |
1817 | #define I_CLEAR (1 << 6) | |
eaff8079 | 1818 | #define __I_SYNC 7 |
1c0eeaf5 | 1819 | #define I_SYNC (1 << __I_SYNC) |
9e38d86f | 1820 | #define I_REFERENCED (1 << 8) |
bd5fe6c5 | 1821 | #define __I_DIO_WAKEUP 9 |
ac74d8d6 | 1822 | #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) |
f4e0c30c | 1823 | #define I_LINKABLE (1 << 10) |
0ae45f63 TT |
1824 | #define I_DIRTY_TIME (1 << 11) |
1825 | #define __I_DIRTY_TIME_EXPIRED 12 | |
1826 | #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED) | |
1da177e4 LT |
1827 | |
1828 | #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) | |
0ae45f63 | 1829 | #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) |
1da177e4 LT |
1830 | |
1831 | extern void __mark_inode_dirty(struct inode *, int); | |
1832 | static inline void mark_inode_dirty(struct inode *inode) | |
1833 | { | |
1834 | __mark_inode_dirty(inode, I_DIRTY); | |
1835 | } | |
1836 | ||
1837 | static inline void mark_inode_dirty_sync(struct inode *inode) | |
1838 | { | |
1839 | __mark_inode_dirty(inode, I_DIRTY_SYNC); | |
1840 | } | |
1841 | ||
7ada4db8 MS |
1842 | extern void inc_nlink(struct inode *inode); |
1843 | extern void drop_nlink(struct inode *inode); | |
1844 | extern void clear_nlink(struct inode *inode); | |
1845 | extern void set_nlink(struct inode *inode, unsigned int nlink); | |
d8c76e6f DH |
1846 | |
1847 | static inline void inode_inc_link_count(struct inode *inode) | |
1848 | { | |
1849 | inc_nlink(inode); | |
66d21730 AD |
1850 | mark_inode_dirty(inode); |
1851 | } | |
1852 | ||
9a53c3a7 DH |
1853 | static inline void inode_dec_link_count(struct inode *inode) |
1854 | { | |
1855 | drop_nlink(inode); | |
66d21730 AD |
1856 | mark_inode_dirty(inode); |
1857 | } | |
1858 | ||
25ec56b5 JNC |
1859 | /** |
1860 | * inode_inc_iversion - increments i_version | |
1861 | * @inode: inode that need to be updated | |
1862 | * | |
1863 | * Every time the inode is modified, the i_version field will be incremented. | |
1864 | * The filesystem has to be mounted with i_version flag | |
1865 | */ | |
1866 | ||
1867 | static inline void inode_inc_iversion(struct inode *inode) | |
1868 | { | |
1869 | spin_lock(&inode->i_lock); | |
1870 | inode->i_version++; | |
1871 | spin_unlock(&inode->i_lock); | |
1872 | } | |
1873 | ||
c3b2da31 JB |
1874 | enum file_time_flags { |
1875 | S_ATIME = 1, | |
1876 | S_MTIME = 2, | |
1877 | S_CTIME = 4, | |
1878 | S_VERSION = 8, | |
1879 | }; | |
1880 | ||
8fa9dd24 | 1881 | extern bool atime_needs_update(const struct path *, struct inode *); |
badcf2b7 | 1882 | extern void touch_atime(const struct path *); |
1da177e4 LT |
1883 | static inline void file_accessed(struct file *file) |
1884 | { | |
1885 | if (!(file->f_flags & O_NOATIME)) | |
68ac1234 | 1886 | touch_atime(&file->f_path); |
1da177e4 LT |
1887 | } |
1888 | ||
1889 | int sync_inode(struct inode *inode, struct writeback_control *wbc); | |
c3765016 | 1890 | int sync_inode_metadata(struct inode *inode, int wait); |
1da177e4 | 1891 | |
1da177e4 LT |
1892 | struct file_system_type { |
1893 | const char *name; | |
1894 | int fs_flags; | |
bbc1096a DH |
1895 | #define FS_REQUIRES_DEV 1 |
1896 | #define FS_BINARY_MOUNTDATA 2 | |
1897 | #define FS_HAS_SUBTYPE 4 | |
0c55cfc4 EB |
1898 | #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ |
1899 | #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */ | |
bbc1096a | 1900 | #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ |
c96e41e9 AV |
1901 | struct dentry *(*mount) (struct file_system_type *, int, |
1902 | const char *, void *); | |
1da177e4 LT |
1903 | void (*kill_sb) (struct super_block *); |
1904 | struct module *owner; | |
1905 | struct file_system_type * next; | |
a5166169 | 1906 | struct hlist_head fs_supers; |
d475fd42 | 1907 | |
cf516249 | 1908 | struct lock_class_key s_lock_key; |
897c6ff9 | 1909 | struct lock_class_key s_umount_key; |
51ee049e | 1910 | struct lock_class_key s_vfs_rename_key; |
5accdf82 | 1911 | struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; |
d475fd42 PZ |
1912 | |
1913 | struct lock_class_key i_lock_key; | |
1914 | struct lock_class_key i_mutex_key; | |
14358e6d | 1915 | struct lock_class_key i_mutex_dir_key; |
1da177e4 LT |
1916 | }; |
1917 | ||
7f78e035 EB |
1918 | #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) |
1919 | ||
ceefda69 AV |
1920 | extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags, |
1921 | void *data, int (*fill_super)(struct super_block *, void *, int)); | |
152a0836 AV |
1922 | extern struct dentry *mount_bdev(struct file_system_type *fs_type, |
1923 | int flags, const char *dev_name, void *data, | |
1924 | int (*fill_super)(struct super_block *, void *, int)); | |
fc14f2fe AV |
1925 | extern struct dentry *mount_single(struct file_system_type *fs_type, |
1926 | int flags, void *data, | |
1927 | int (*fill_super)(struct super_block *, void *, int)); | |
3c26ff6e AV |
1928 | extern struct dentry *mount_nodev(struct file_system_type *fs_type, |
1929 | int flags, void *data, | |
1930 | int (*fill_super)(struct super_block *, void *, int)); | |
ea441d11 | 1931 | extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); |
1da177e4 LT |
1932 | void generic_shutdown_super(struct super_block *sb); |
1933 | void kill_block_super(struct super_block *sb); | |
1934 | void kill_anon_super(struct super_block *sb); | |
1935 | void kill_litter_super(struct super_block *sb); | |
1936 | void deactivate_super(struct super_block *sb); | |
74dbbdd7 | 1937 | void deactivate_locked_super(struct super_block *sb); |
1da177e4 | 1938 | int set_anon_super(struct super_block *s, void *data); |
0ee5dc67 AV |
1939 | int get_anon_bdev(dev_t *); |
1940 | void free_anon_bdev(dev_t); | |
1da177e4 LT |
1941 | struct super_block *sget(struct file_system_type *type, |
1942 | int (*test)(struct super_block *,void *), | |
1943 | int (*set)(struct super_block *,void *), | |
9249e17f | 1944 | int flags, void *data); |
51139ada | 1945 | extern struct dentry *mount_pseudo(struct file_system_type *, char *, |
c74a1cbb AV |
1946 | const struct super_operations *ops, |
1947 | const struct dentry_operations *dops, | |
1948 | unsigned long); | |
1da177e4 LT |
1949 | |
1950 | /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ | |
1951 | #define fops_get(fops) \ | |
1952 | (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) | |
1953 | #define fops_put(fops) \ | |
1954 | do { if (fops) module_put((fops)->owner); } while(0) | |
e84f9e57 AV |
1955 | /* |
1956 | * This one is to be used *ONLY* from ->open() instances. | |
1957 | * fops must be non-NULL, pinned down *and* module dependencies | |
1958 | * should be sufficient to pin the caller down as well. | |
1959 | */ | |
1960 | #define replace_fops(f, fops) \ | |
1961 | do { \ | |
1962 | struct file *__file = (f); \ | |
1963 | fops_put(__file->f_op); \ | |
1964 | BUG_ON(!(__file->f_op = (fops))); \ | |
1965 | } while(0) | |
1da177e4 LT |
1966 | |
1967 | extern int register_filesystem(struct file_system_type *); | |
1968 | extern int unregister_filesystem(struct file_system_type *); | |
8bf9725c PE |
1969 | extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data); |
1970 | #define kern_mount(type) kern_mount_data(type, NULL) | |
423e0ab0 | 1971 | extern void kern_unmount(struct vfsmount *mnt); |
1da177e4 LT |
1972 | extern int may_umount_tree(struct vfsmount *); |
1973 | extern int may_umount(struct vfsmount *); | |
5e6123f3 SL |
1974 | extern long do_mount(const char *, const char __user *, |
1975 | const char *, unsigned long, void *); | |
589ff870 | 1976 | extern struct vfsmount *collect_mounts(struct path *); |
8aec0809 | 1977 | extern void drop_collected_mounts(struct vfsmount *); |
1f707137 AV |
1978 | extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *, |
1979 | struct vfsmount *); | |
ebabe9a9 | 1980 | extern int vfs_statfs(struct path *, struct kstatfs *); |
c8b91acc AV |
1981 | extern int user_statfs(const char __user *, struct kstatfs *); |
1982 | extern int fd_statfs(int, struct kstatfs *); | |
cf31e70d | 1983 | extern int vfs_ustat(dev_t, struct kstatfs *); |
18e9e510 JB |
1984 | extern int freeze_super(struct super_block *super); |
1985 | extern int thaw_super(struct super_block *super); | |
02125a82 | 1986 | extern bool our_mnt(struct vfsmount *mnt); |
e51db735 | 1987 | extern bool fs_fully_visible(struct file_system_type *); |
1da177e4 | 1988 | |
ce3b0f8d AV |
1989 | extern int current_umask(void); |
1990 | ||
5a14696c BF |
1991 | extern void ihold(struct inode * inode); |
1992 | extern void iput(struct inode *); | |
0ae45f63 | 1993 | extern int generic_update_time(struct inode *, struct timespec *, int); |
5a14696c | 1994 | |
1fa97e8b | 1995 | static inline struct inode *file_inode(const struct file *f) |
d7a06983 JL |
1996 | { |
1997 | return f->f_inode; | |
1998 | } | |
1999 | ||
f87fd4c2 | 2000 | /* /sys/fs */ |
00d26666 | 2001 | extern struct kobject *fs_kobj; |
f87fd4c2 | 2002 | |
435f49a5 | 2003 | #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK) |
bfcd17a6 | 2004 | |
1da177e4 LT |
2005 | #define FLOCK_VERIFY_READ 1 |
2006 | #define FLOCK_VERIFY_WRITE 2 | |
2007 | ||
bfcd17a6 | 2008 | #ifdef CONFIG_FILE_LOCKING |
d7a06983 | 2009 | extern int locks_mandatory_locked(struct file *); |
1da177e4 LT |
2010 | extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t); |
2011 | ||
2012 | /* | |
2013 | * Candidates for mandatory locking have the setgid bit set | |
2014 | * but no group execute bit - an otherwise meaningless combination. | |
2015 | */ | |
a16877ca PE |
2016 | |
2017 | static inline int __mandatory_lock(struct inode *ino) | |
2018 | { | |
2019 | return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; | |
2020 | } | |
2021 | ||
2022 | /* | |
2023 | * ... and these candidates should be on MS_MANDLOCK mounted fs, | |
2024 | * otherwise these will be advisory locks | |
2025 | */ | |
2026 | ||
2027 | static inline int mandatory_lock(struct inode *ino) | |
2028 | { | |
2029 | return IS_MANDLOCK(ino) && __mandatory_lock(ino); | |
2030 | } | |
1da177e4 | 2031 | |
d7a06983 | 2032 | static inline int locks_verify_locked(struct file *file) |
1da177e4 | 2033 | { |
d7a06983 JL |
2034 | if (mandatory_lock(file_inode(file))) |
2035 | return locks_mandatory_locked(file); | |
1da177e4 LT |
2036 | return 0; |
2037 | } | |
2038 | ||
1da177e4 LT |
2039 | static inline int locks_verify_truncate(struct inode *inode, |
2040 | struct file *filp, | |
2041 | loff_t size) | |
2042 | { | |
bd61e0a9 | 2043 | if (inode->i_flctx && mandatory_lock(inode)) |
1da177e4 LT |
2044 | return locks_mandatory_area( |
2045 | FLOCK_VERIFY_WRITE, inode, filp, | |
2046 | size < inode->i_size ? size : inode->i_size, | |
2047 | (size < inode->i_size ? inode->i_size - size | |
2048 | : size - inode->i_size) | |
2049 | ); | |
2050 | return 0; | |
2051 | } | |
2052 | ||
2053 | static inline int break_lease(struct inode *inode, unsigned int mode) | |
2054 | { | |
24cbe784 JL |
2055 | /* |
2056 | * Since this check is lockless, we must ensure that any refcounts | |
8116bf4c JL |
2057 | * taken are done before checking i_flctx->flc_lease. Otherwise, we |
2058 | * could end up racing with tasks trying to set a new lease on this | |
2059 | * file. | |
24cbe784 JL |
2060 | */ |
2061 | smp_mb(); | |
8634b51f | 2062 | if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) |
df4e8d2c | 2063 | return __break_lease(inode, mode, FL_LEASE); |
1da177e4 LT |
2064 | return 0; |
2065 | } | |
df4e8d2c BF |
2066 | |
2067 | static inline int break_deleg(struct inode *inode, unsigned int mode) | |
2068 | { | |
962bd40b JL |
2069 | /* |
2070 | * Since this check is lockless, we must ensure that any refcounts | |
8116bf4c JL |
2071 | * taken are done before checking i_flctx->flc_lease. Otherwise, we |
2072 | * could end up racing with tasks trying to set a new lease on this | |
2073 | * file. | |
962bd40b JL |
2074 | */ |
2075 | smp_mb(); | |
8634b51f | 2076 | if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) |
df4e8d2c | 2077 | return __break_lease(inode, mode, FL_DELEG); |
1da177e4 LT |
2078 | return 0; |
2079 | } | |
df4e8d2c | 2080 | |
5a14696c BF |
2081 | static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) |
2082 | { | |
2083 | int ret; | |
2084 | ||
2085 | ret = break_deleg(inode, O_WRONLY|O_NONBLOCK); | |
2086 | if (ret == -EWOULDBLOCK && delegated_inode) { | |
2087 | *delegated_inode = inode; | |
2088 | ihold(inode); | |
2089 | } | |
2090 | return ret; | |
2091 | } | |
2092 | ||
2093 | static inline int break_deleg_wait(struct inode **delegated_inode) | |
2094 | { | |
2095 | int ret; | |
2096 | ||
2097 | ret = break_deleg(*delegated_inode, O_WRONLY); | |
2098 | iput(*delegated_inode); | |
2099 | *delegated_inode = NULL; | |
2100 | return ret; | |
2101 | } | |
2102 | ||
11afe9f7 CH |
2103 | static inline int break_layout(struct inode *inode, bool wait) |
2104 | { | |
2105 | smp_mb(); | |
2106 | if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) | |
2107 | return __break_lease(inode, | |
2108 | wait ? O_WRONLY : O_WRONLY | O_NONBLOCK, | |
2109 | FL_LAYOUT); | |
2110 | return 0; | |
2111 | } | |
2112 | ||
bfcd17a6 | 2113 | #else /* !CONFIG_FILE_LOCKING */ |
d7a06983 | 2114 | static inline int locks_mandatory_locked(struct file *file) |
af5df566 SW |
2115 | { |
2116 | return 0; | |
2117 | } | |
2118 | ||
2119 | static inline int locks_mandatory_area(int rw, struct inode *inode, | |
2120 | struct file *filp, loff_t offset, | |
2121 | size_t count) | |
2122 | { | |
2123 | return 0; | |
2124 | } | |
2125 | ||
2126 | static inline int __mandatory_lock(struct inode *inode) | |
2127 | { | |
2128 | return 0; | |
2129 | } | |
2130 | ||
2131 | static inline int mandatory_lock(struct inode *inode) | |
2132 | { | |
2133 | return 0; | |
2134 | } | |
2135 | ||
d7a06983 | 2136 | static inline int locks_verify_locked(struct file *file) |
af5df566 SW |
2137 | { |
2138 | return 0; | |
2139 | } | |
2140 | ||
2141 | static inline int locks_verify_truncate(struct inode *inode, struct file *filp, | |
2142 | size_t size) | |
2143 | { | |
2144 | return 0; | |
2145 | } | |
2146 | ||
2147 | static inline int break_lease(struct inode *inode, unsigned int mode) | |
2148 | { | |
2149 | return 0; | |
2150 | } | |
2151 | ||
df4e8d2c BF |
2152 | static inline int break_deleg(struct inode *inode, unsigned int mode) |
2153 | { | |
2154 | return 0; | |
2155 | } | |
5a14696c BF |
2156 | |
2157 | static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) | |
2158 | { | |
2159 | return 0; | |
2160 | } | |
2161 | ||
2162 | static inline int break_deleg_wait(struct inode **delegated_inode) | |
2163 | { | |
2164 | BUG(); | |
2165 | return 0; | |
2166 | } | |
2167 | ||
11afe9f7 CH |
2168 | static inline int break_layout(struct inode *inode, bool wait) |
2169 | { | |
2170 | return 0; | |
2171 | } | |
2172 | ||
bfcd17a6 | 2173 | #endif /* CONFIG_FILE_LOCKING */ |
1da177e4 LT |
2174 | |
2175 | /* fs/open.c */ | |
adb5c247 | 2176 | struct audit_names; |
91a27b2a | 2177 | struct filename { |
adb5c247 JL |
2178 | const char *name; /* pointer to actual string */ |
2179 | const __user char *uptr; /* original userland pointer */ | |
2180 | struct audit_names *aname; | |
55422d0b | 2181 | int refcnt; |
fd2f7cb5 | 2182 | const char iname[]; |
91a27b2a | 2183 | }; |
1da177e4 | 2184 | |
a02de960 | 2185 | extern long vfs_truncate(struct path *, loff_t); |
4a30131e N |
2186 | extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, |
2187 | struct file *filp); | |
72c72bdf | 2188 | extern int vfs_fallocate(struct file *file, int mode, loff_t offset, |
3e63cbb1 | 2189 | loff_t len); |
8e8a1407 | 2190 | extern long do_sys_open(int dfd, const char __user *filename, int flags, |
a218d0fd | 2191 | umode_t mode); |
669abf4e | 2192 | extern struct file *file_open_name(struct filename *, int, umode_t); |
a218d0fd | 2193 | extern struct file *filp_open(const char *, int, umode_t); |
73d049a4 AV |
2194 | extern struct file *file_open_root(struct dentry *, struct vfsmount *, |
2195 | const char *, int); | |
765927b2 | 2196 | extern struct file * dentry_open(const struct path *, int, const struct cred *); |
1da177e4 | 2197 | extern int filp_close(struct file *, fl_owner_t id); |
91a27b2a | 2198 | |
51f39a1f | 2199 | extern struct filename *getname_flags(const char __user *, int, int *); |
91a27b2a | 2200 | extern struct filename *getname(const char __user *); |
c4ad8f98 | 2201 | extern struct filename *getname_kernel(const char *); |
55422d0b | 2202 | extern void putname(struct filename *name); |
91a27b2a | 2203 | |
47237687 AV |
2204 | enum { |
2205 | FILE_CREATED = 1, | |
2206 | FILE_OPENED = 2 | |
2207 | }; | |
30d90494 AV |
2208 | extern int finish_open(struct file *file, struct dentry *dentry, |
2209 | int (*open)(struct inode *, struct file *), | |
2210 | int *opened); | |
e45198a6 | 2211 | extern int finish_no_open(struct file *file, struct dentry *dentry); |
1da177e4 | 2212 | |
3e63cbb1 AJ |
2213 | /* fs/ioctl.c */ |
2214 | ||
2215 | extern int ioctl_preallocate(struct file *filp, void __user *argp); | |
2216 | ||
1da177e4 LT |
2217 | /* fs/dcache.c */ |
2218 | extern void __init vfs_caches_init_early(void); | |
2219 | extern void __init vfs_caches_init(unsigned long); | |
2220 | ||
b86c089b CL |
2221 | extern struct kmem_cache *names_cachep; |
2222 | ||
a608ca21 | 2223 | #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) |
3446a8aa | 2224 | #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) |
1da177e4 | 2225 | |
9361401e | 2226 | #ifdef CONFIG_BLOCK |
1da177e4 | 2227 | extern int register_blkdev(unsigned int, const char *); |
f4480240 | 2228 | extern void unregister_blkdev(unsigned int, const char *); |
1da177e4 | 2229 | extern struct block_device *bdget(dev_t); |
dddac6a7 | 2230 | extern struct block_device *bdgrab(struct block_device *bdev); |
1da177e4 LT |
2231 | extern void bd_set_size(struct block_device *, loff_t size); |
2232 | extern void bd_forget(struct inode *inode); | |
2233 | extern void bdput(struct block_device *); | |
585d3bc0 | 2234 | extern void invalidate_bdev(struct block_device *); |
5c0d6b60 | 2235 | extern void iterate_bdevs(void (*)(struct block_device *, void *), void *); |
585d3bc0 | 2236 | extern int sync_blockdev(struct block_device *bdev); |
ff01bb48 | 2237 | extern void kill_bdev(struct block_device *); |
585d3bc0 | 2238 | extern struct super_block *freeze_bdev(struct block_device *); |
c2d75438 | 2239 | extern void emergency_thaw_all(void); |
585d3bc0 NP |
2240 | extern int thaw_bdev(struct block_device *bdev, struct super_block *sb); |
2241 | extern int fsync_bdev(struct block_device *); | |
a8855990 | 2242 | extern int sb_is_blkdev_sb(struct super_block *sb); |
9361401e DH |
2243 | #else |
2244 | static inline void bd_forget(struct inode *inode) {} | |
47e4491b | 2245 | static inline int sync_blockdev(struct block_device *bdev) { return 0; } |
ff01bb48 | 2246 | static inline void kill_bdev(struct block_device *bdev) {} |
47e4491b AV |
2247 | static inline void invalidate_bdev(struct block_device *bdev) {} |
2248 | ||
2249 | static inline struct super_block *freeze_bdev(struct block_device *sb) | |
2250 | { | |
2251 | return NULL; | |
2252 | } | |
2253 | ||
2254 | static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb) | |
2255 | { | |
2256 | return 0; | |
2257 | } | |
5c0d6b60 JK |
2258 | |
2259 | static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg) | |
2260 | { | |
2261 | } | |
a8855990 JK |
2262 | |
2263 | static inline int sb_is_blkdev_sb(struct super_block *sb) | |
2264 | { | |
2265 | return 0; | |
2266 | } | |
9361401e | 2267 | #endif |
60b0680f | 2268 | extern int sync_filesystem(struct super_block *); |
9361401e | 2269 | extern const struct file_operations def_blk_fops; |
4b6f5d20 | 2270 | extern const struct file_operations def_chr_fops; |
9361401e | 2271 | #ifdef CONFIG_BLOCK |
1da177e4 | 2272 | extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); |
56b26add | 2273 | extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); |
1da177e4 | 2274 | extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); |
e525fd89 | 2275 | extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder); |
d4d77629 TH |
2276 | extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode, |
2277 | void *holder); | |
2278 | extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, | |
2279 | void *holder); | |
4385bab1 | 2280 | extern void blkdev_put(struct block_device *bdev, fmode_t mode); |
641dc636 | 2281 | #ifdef CONFIG_SYSFS |
e09b457b | 2282 | extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk); |
49731baa TH |
2283 | extern void bd_unlink_disk_holder(struct block_device *bdev, |
2284 | struct gendisk *disk); | |
641dc636 | 2285 | #else |
e09b457b TH |
2286 | static inline int bd_link_disk_holder(struct block_device *bdev, |
2287 | struct gendisk *disk) | |
2288 | { | |
2289 | return 0; | |
2290 | } | |
49731baa TH |
2291 | static inline void bd_unlink_disk_holder(struct block_device *bdev, |
2292 | struct gendisk *disk) | |
2293 | { | |
2294 | } | |
641dc636 | 2295 | #endif |
9361401e | 2296 | #endif |
1da177e4 LT |
2297 | |
2298 | /* fs/char_dev.c */ | |
68eef3b4 | 2299 | #define CHRDEV_MAJOR_HASH_SIZE 255 |
1da177e4 LT |
2300 | extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); |
2301 | extern int register_chrdev_region(dev_t, unsigned, const char *); | |
1905b1bf TH |
2302 | extern int __register_chrdev(unsigned int major, unsigned int baseminor, |
2303 | unsigned int count, const char *name, | |
2304 | const struct file_operations *fops); | |
2305 | extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, | |
2306 | unsigned int count, const char *name); | |
1da177e4 | 2307 | extern void unregister_chrdev_region(dev_t, unsigned); |
68eef3b4 | 2308 | extern void chrdev_show(struct seq_file *,off_t); |
1da177e4 | 2309 | |
1905b1bf TH |
2310 | static inline int register_chrdev(unsigned int major, const char *name, |
2311 | const struct file_operations *fops) | |
2312 | { | |
2313 | return __register_chrdev(major, 0, 256, name, fops); | |
2314 | } | |
2315 | ||
2316 | static inline void unregister_chrdev(unsigned int major, const char *name) | |
2317 | { | |
2318 | __unregister_chrdev(major, 0, 256, name); | |
2319 | } | |
2320 | ||
1da177e4 LT |
2321 | /* fs/block_dev.c */ |
2322 | #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ | |
1f014290 | 2323 | #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ |
9361401e DH |
2324 | |
2325 | #ifdef CONFIG_BLOCK | |
2326 | #define BLKDEV_MAJOR_HASH_SIZE 255 | |
1da177e4 LT |
2327 | extern const char *__bdevname(dev_t, char *buffer); |
2328 | extern const char *bdevname(struct block_device *bdev, char *buffer); | |
2329 | extern struct block_device *lookup_bdev(const char *); | |
68eef3b4 | 2330 | extern void blkdev_show(struct seq_file *,off_t); |
30c40d2c | 2331 | |
9361401e DH |
2332 | #else |
2333 | #define BLKDEV_MAJOR_HASH_SIZE 0 | |
2334 | #endif | |
1da177e4 LT |
2335 | |
2336 | extern void init_special_inode(struct inode *, umode_t, dev_t); | |
2337 | ||
2338 | /* Invalid inode operations -- fs/bad_inode.c */ | |
2339 | extern void make_bad_inode(struct inode *); | |
2340 | extern int is_bad_inode(struct inode *); | |
2341 | ||
9361401e | 2342 | #ifdef CONFIG_BLOCK |
1da177e4 LT |
2343 | /* |
2344 | * return READ, READA, or WRITE | |
2345 | */ | |
2346 | #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK)) | |
2347 | ||
2348 | /* | |
2349 | * return data direction, READ or WRITE | |
2350 | */ | |
2351 | #define bio_data_dir(bio) ((bio)->bi_rw & 1) | |
2352 | ||
c3279d14 AP |
2353 | extern void check_disk_size_change(struct gendisk *disk, |
2354 | struct block_device *bdev); | |
0c002c2f | 2355 | extern int revalidate_disk(struct gendisk *); |
1da177e4 | 2356 | extern int check_disk_change(struct block_device *); |
93b270f7 | 2357 | extern int __invalidate_device(struct block_device *, bool); |
1da177e4 | 2358 | extern int invalidate_partition(struct gendisk *, int); |
9361401e | 2359 | #endif |
1da177e4 LT |
2360 | unsigned long invalidate_mapping_pages(struct address_space *mapping, |
2361 | pgoff_t start, pgoff_t end); | |
54bc4855 | 2362 | |
1da177e4 LT |
2363 | static inline void invalidate_remote_inode(struct inode *inode) |
2364 | { | |
2365 | if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || | |
2366 | S_ISLNK(inode->i_mode)) | |
fc0ecff6 | 2367 | invalidate_mapping_pages(inode->i_mapping, 0, -1); |
1da177e4 LT |
2368 | } |
2369 | extern int invalidate_inode_pages2(struct address_space *mapping); | |
2370 | extern int invalidate_inode_pages2_range(struct address_space *mapping, | |
2371 | pgoff_t start, pgoff_t end); | |
2372 | extern int write_inode_now(struct inode *, int); | |
2373 | extern int filemap_fdatawrite(struct address_space *); | |
2374 | extern int filemap_flush(struct address_space *); | |
2375 | extern int filemap_fdatawait(struct address_space *); | |
d3bccb6f JK |
2376 | extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, |
2377 | loff_t lend); | |
1da177e4 LT |
2378 | extern int filemap_write_and_wait(struct address_space *mapping); |
2379 | extern int filemap_write_and_wait_range(struct address_space *mapping, | |
2380 | loff_t lstart, loff_t lend); | |
ebcf28e1 AM |
2381 | extern int __filemap_fdatawrite_range(struct address_space *mapping, |
2382 | loff_t start, loff_t end, int sync_mode); | |
f4c0a0fd JK |
2383 | extern int filemap_fdatawrite_range(struct address_space *mapping, |
2384 | loff_t start, loff_t end); | |
ebcf28e1 | 2385 | |
8018ab05 CH |
2386 | extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, |
2387 | int datasync); | |
2388 | extern int vfs_fsync(struct file *file, int datasync); | |
d311d79d AV |
2389 | static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count) |
2390 | { | |
2391 | if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host)) | |
2392 | return 0; | |
2393 | return vfs_fsync_range(file, pos, pos + count - 1, | |
2394 | (file->f_flags & __O_SYNC) ? 0 : 1); | |
2395 | } | |
1da177e4 LT |
2396 | extern void emergency_sync(void); |
2397 | extern void emergency_remount(void); | |
9361401e | 2398 | #ifdef CONFIG_BLOCK |
1da177e4 | 2399 | extern sector_t bmap(struct inode *, sector_t); |
9361401e | 2400 | #endif |
27ac0ffe | 2401 | extern int notify_change(struct dentry *, struct iattr *, struct inode **); |
f419a2e3 | 2402 | extern int inode_permission(struct inode *, int); |
bd5d0856 | 2403 | extern int __inode_permission(struct inode *, int); |
2830ba7f | 2404 | extern int generic_permission(struct inode *, int); |
cbdf35bc | 2405 | extern int __check_sticky(struct inode *dir, struct inode *inode); |
1da177e4 | 2406 | |
f696a365 MS |
2407 | static inline bool execute_ok(struct inode *inode) |
2408 | { | |
2409 | return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); | |
2410 | } | |
2411 | ||
8d71db4f AV |
2412 | static inline void file_start_write(struct file *file) |
2413 | { | |
2414 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2415 | return; | |
2416 | __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true); | |
2417 | } | |
2418 | ||
5ae98f15 JK |
2419 | static inline bool file_start_write_trylock(struct file *file) |
2420 | { | |
2421 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2422 | return true; | |
2423 | return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false); | |
2424 | } | |
2425 | ||
8d71db4f AV |
2426 | static inline void file_end_write(struct file *file) |
2427 | { | |
2428 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2429 | return; | |
2430 | __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE); | |
2431 | } | |
2432 | ||
07b8ce1e AV |
2433 | /* |
2434 | * get_write_access() gets write permission for a file. | |
2435 | * put_write_access() releases this write permission. | |
2436 | * This is used for regular files. | |
2437 | * We cannot support write (and maybe mmap read-write shared) accesses and | |
2438 | * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode | |
2439 | * can have the following values: | |
2440 | * 0: no writers, no VM_DENYWRITE mappings | |
2441 | * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist | |
2442 | * > 0: (i_writecount) users are writing to the file. | |
2443 | * | |
2444 | * Normally we operate on that counter with atomic_{inc,dec} and it's safe | |
2445 | * except for the cases where we don't hold i_writecount yet. Then we need to | |
2446 | * use {get,deny}_write_access() - these functions check the sign and refuse | |
2447 | * to do the change if sign is wrong. | |
2448 | */ | |
2449 | static inline int get_write_access(struct inode *inode) | |
2450 | { | |
2451 | return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; | |
2452 | } | |
2453 | static inline int deny_write_access(struct file *file) | |
2454 | { | |
496ad9aa | 2455 | struct inode *inode = file_inode(file); |
07b8ce1e AV |
2456 | return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; |
2457 | } | |
1da177e4 LT |
2458 | static inline void put_write_access(struct inode * inode) |
2459 | { | |
2460 | atomic_dec(&inode->i_writecount); | |
2461 | } | |
2462 | static inline void allow_write_access(struct file *file) | |
2463 | { | |
2464 | if (file) | |
496ad9aa | 2465 | atomic_inc(&file_inode(file)->i_writecount); |
1da177e4 | 2466 | } |
f1fe29b4 DH |
2467 | static inline bool inode_is_open_for_write(const struct inode *inode) |
2468 | { | |
2469 | return atomic_read(&inode->i_writecount) > 0; | |
2470 | } | |
2471 | ||
a5c96ebf MZ |
2472 | #ifdef CONFIG_IMA |
2473 | static inline void i_readcount_dec(struct inode *inode) | |
2474 | { | |
2475 | BUG_ON(!atomic_read(&inode->i_readcount)); | |
2476 | atomic_dec(&inode->i_readcount); | |
2477 | } | |
2478 | static inline void i_readcount_inc(struct inode *inode) | |
2479 | { | |
2480 | atomic_inc(&inode->i_readcount); | |
2481 | } | |
2482 | #else | |
2483 | static inline void i_readcount_dec(struct inode *inode) | |
2484 | { | |
2485 | return; | |
2486 | } | |
2487 | static inline void i_readcount_inc(struct inode *inode) | |
2488 | { | |
2489 | return; | |
2490 | } | |
2491 | #endif | |
ed8cae8b | 2492 | extern int do_pipe_flags(int *, int); |
1da177e4 | 2493 | |
6777d773 | 2494 | extern int kernel_read(struct file *, loff_t, char *, unsigned long); |
7bb307e8 | 2495 | extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t); |
ed44724b | 2496 | extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *); |
1da177e4 LT |
2497 | extern struct file * open_exec(const char *); |
2498 | ||
2499 | /* fs/dcache.c -- generic fs support functions */ | |
2500 | extern int is_subdir(struct dentry *, struct dentry *); | |
2096f759 | 2501 | extern int path_is_under(struct path *, struct path *); |
1da177e4 | 2502 | |
9bf39ab2 MS |
2503 | extern char *file_path(struct file *, char *, int); |
2504 | ||
1da177e4 LT |
2505 | #include <linux/err.h> |
2506 | ||
2507 | /* needed for stackable file system support */ | |
965c8e59 | 2508 | extern loff_t default_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 2509 | |
965c8e59 | 2510 | extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 2511 | |
54e34621 | 2512 | extern int inode_init_always(struct super_block *, struct inode *); |
1da177e4 | 2513 | extern void inode_init_once(struct inode *); |
2aa15890 | 2514 | extern void address_space_init_once(struct address_space *mapping); |
1da177e4 LT |
2515 | extern struct inode * igrab(struct inode *); |
2516 | extern ino_t iunique(struct super_block *, ino_t); | |
2517 | extern int inode_needs_sync(struct inode *inode); | |
45321ac5 | 2518 | extern int generic_delete_inode(struct inode *inode); |
9bcb4b73 AV |
2519 | static inline int generic_drop_inode(struct inode *inode) |
2520 | { | |
2521 | return !inode->i_nlink || inode_unhashed(inode); | |
2522 | } | |
1da177e4 | 2523 | |
88bd5121 AA |
2524 | extern struct inode *ilookup5_nowait(struct super_block *sb, |
2525 | unsigned long hashval, int (*test)(struct inode *, void *), | |
2526 | void *data); | |
1da177e4 LT |
2527 | extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, |
2528 | int (*test)(struct inode *, void *), void *data); | |
2529 | extern struct inode *ilookup(struct super_block *sb, unsigned long ino); | |
2530 | ||
2531 | extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); | |
2532 | extern struct inode * iget_locked(struct super_block *, unsigned long); | |
fe032c42 TT |
2533 | extern struct inode *find_inode_nowait(struct super_block *, |
2534 | unsigned long, | |
2535 | int (*match)(struct inode *, | |
2536 | unsigned long, void *), | |
2537 | void *data); | |
261bca86 AV |
2538 | extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); |
2539 | extern int insert_inode_locked(struct inode *); | |
e096d0c7 JB |
2540 | #ifdef CONFIG_DEBUG_LOCK_ALLOC |
2541 | extern void lockdep_annotate_inode_mutex_key(struct inode *inode); | |
2542 | #else | |
2543 | static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; | |
2544 | #endif | |
1da177e4 | 2545 | extern void unlock_new_inode(struct inode *); |
85fe4025 | 2546 | extern unsigned int get_next_ino(void); |
1da177e4 | 2547 | |
1da177e4 | 2548 | extern void __iget(struct inode * inode); |
b46980fe | 2549 | extern void iget_failed(struct inode *); |
dbd5768f | 2550 | extern void clear_inode(struct inode *); |
2e00c97e | 2551 | extern void __destroy_inode(struct inode *); |
a209dfc7 ED |
2552 | extern struct inode *new_inode_pseudo(struct super_block *sb); |
2553 | extern struct inode *new_inode(struct super_block *sb); | |
ff0c7d15 | 2554 | extern void free_inode_nonrcu(struct inode *inode); |
01de85e0 | 2555 | extern int should_remove_suid(struct dentry *); |
2f1936b8 | 2556 | extern int file_remove_suid(struct file *); |
1da177e4 LT |
2557 | |
2558 | extern void __insert_inode_hash(struct inode *, unsigned long hashval); | |
646ec461 CH |
2559 | static inline void insert_inode_hash(struct inode *inode) |
2560 | { | |
1da177e4 LT |
2561 | __insert_inode_hash(inode, inode->i_ino); |
2562 | } | |
f2ee7abf ED |
2563 | |
2564 | extern void __remove_inode_hash(struct inode *); | |
2565 | static inline void remove_inode_hash(struct inode *inode) | |
2566 | { | |
2567 | if (!inode_unhashed(inode)) | |
2568 | __remove_inode_hash(inode); | |
2569 | } | |
2570 | ||
646ec461 | 2571 | extern void inode_sb_list_add(struct inode *inode); |
1da177e4 | 2572 | |
9361401e | 2573 | #ifdef CONFIG_BLOCK |
1da177e4 LT |
2574 | extern void submit_bio(int, struct bio *); |
2575 | extern int bdev_read_only(struct block_device *); | |
9361401e | 2576 | #endif |
1da177e4 LT |
2577 | extern int set_blocksize(struct block_device *, int); |
2578 | extern int sb_set_blocksize(struct super_block *, int); | |
2579 | extern int sb_min_blocksize(struct super_block *, int); | |
2580 | ||
2581 | extern int generic_file_mmap(struct file *, struct vm_area_struct *); | |
2582 | extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); | |
3309dd04 | 2583 | extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); |
ed978a81 | 2584 | extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); |
8174202b | 2585 | extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); |
8174202b | 2586 | extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); |
0c949334 | 2587 | extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t); |
3b93f911 | 2588 | extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t); |
88e6faef | 2589 | |
dbe4e192 CH |
2590 | ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos); |
2591 | ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos); | |
2592 | ||
eef99380 | 2593 | /* fs/block_dev.c */ |
b2de525f | 2594 | extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to); |
1456c0a8 | 2595 | extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from); |
02c24a82 JB |
2596 | extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end, |
2597 | int datasync); | |
ff01bb48 | 2598 | extern void block_sync_page(struct page *page); |
eef99380 | 2599 | |
88e6faef | 2600 | /* fs/splice.c */ |
cbb7e577 | 2601 | extern ssize_t generic_file_splice_read(struct file *, loff_t *, |
88e6faef | 2602 | struct pipe_inode_info *, size_t, unsigned int); |
6818173b MS |
2603 | extern ssize_t default_file_splice_read(struct file *, loff_t *, |
2604 | struct pipe_inode_info *, size_t, unsigned int); | |
8d020765 | 2605 | extern ssize_t iter_file_splice_write(struct pipe_inode_info *, |
cbb7e577 | 2606 | struct file *, loff_t *, size_t, unsigned int); |
88e6faef | 2607 | extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, |
cbb7e577 | 2608 | struct file *out, loff_t *, size_t len, unsigned int flags); |
1c118596 MS |
2609 | extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, |
2610 | loff_t *opos, size_t len, unsigned int flags); | |
2611 | ||
88e6faef | 2612 | |
1da177e4 LT |
2613 | extern void |
2614 | file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); | |
965c8e59 AM |
2615 | extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); |
2616 | extern loff_t no_llseek(struct file *file, loff_t offset, int whence); | |
46a1c2c7 | 2617 | extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); |
965c8e59 | 2618 | extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); |
5760495a | 2619 | extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, |
965c8e59 | 2620 | int whence, loff_t maxsize, loff_t eof); |
1bf9d14d AV |
2621 | extern loff_t fixed_size_llseek(struct file *file, loff_t offset, |
2622 | int whence, loff_t size); | |
1da177e4 LT |
2623 | extern int generic_file_open(struct inode * inode, struct file * filp); |
2624 | extern int nonseekable_open(struct inode * inode, struct file * filp); | |
2625 | ||
a95cd631 OS |
2626 | ssize_t dax_do_io(struct kiocb *, struct inode *, struct iov_iter *, loff_t, |
2627 | get_block_t, dio_iodone_t, int flags); | |
289c6aed | 2628 | int dax_clear_blocks(struct inode *, sector_t block, long size); |
25726bc1 | 2629 | int dax_zero_page_range(struct inode *, loff_t from, unsigned len, get_block_t); |
4c0ccfef | 2630 | int dax_truncate_page(struct inode *, loff_t from, get_block_t); |
f7ca90b1 | 2631 | int dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t); |
0e3b210c | 2632 | int dax_pfn_mkwrite(struct vm_area_struct *, struct vm_fault *); |
f7ca90b1 | 2633 | #define dax_mkwrite(vma, vmf, gb) dax_fault(vma, vmf, gb) |
d475c634 | 2634 | |
9361401e | 2635 | #ifdef CONFIG_BLOCK |
facd07b0 JB |
2636 | typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode, |
2637 | loff_t file_offset); | |
1da177e4 LT |
2638 | |
2639 | enum { | |
1e431f5c CH |
2640 | /* need locking between buffered and direct access */ |
2641 | DIO_LOCKING = 0x01, | |
2642 | ||
2643 | /* filesystem does not support filling holes */ | |
2644 | DIO_SKIP_HOLES = 0x02, | |
60392573 CH |
2645 | |
2646 | /* filesystem can handle aio writes beyond i_size */ | |
2647 | DIO_ASYNC_EXTEND = 0x04, | |
fe0f07d0 JA |
2648 | |
2649 | /* inode/fs/bdev does not need truncate protection */ | |
2650 | DIO_SKIP_DIO_COUNT = 0x08, | |
1da177e4 LT |
2651 | }; |
2652 | ||
eafdc7d1 CH |
2653 | void dio_end_io(struct bio *bio, int error); |
2654 | ||
17f8c842 OS |
2655 | ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, |
2656 | struct block_device *bdev, struct iov_iter *iter, | |
2657 | loff_t offset, get_block_t get_block, | |
2658 | dio_iodone_t end_io, dio_submit_t submit_io, | |
2659 | int flags); | |
7bb46a67 | 2660 | |
17f8c842 OS |
2661 | static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, |
2662 | struct inode *inode, | |
2663 | struct iov_iter *iter, loff_t offset, | |
2664 | get_block_t get_block) | |
1da177e4 | 2665 | { |
17f8c842 | 2666 | return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, |
31b14039 | 2667 | offset, get_block, NULL, NULL, |
1e431f5c | 2668 | DIO_LOCKING | DIO_SKIP_HOLES); |
1da177e4 | 2669 | } |
9361401e | 2670 | #endif |
1da177e4 | 2671 | |
1d59d61f | 2672 | void inode_dio_wait(struct inode *inode); |
fe0f07d0 JA |
2673 | |
2674 | /* | |
2675 | * inode_dio_begin - signal start of a direct I/O requests | |
2676 | * @inode: inode the direct I/O happens on | |
2677 | * | |
2678 | * This is called once we've finished processing a direct I/O request, | |
2679 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
2680 | */ | |
2681 | static inline void inode_dio_begin(struct inode *inode) | |
2682 | { | |
2683 | atomic_inc(&inode->i_dio_count); | |
2684 | } | |
2685 | ||
2686 | /* | |
2687 | * inode_dio_end - signal finish of a direct I/O requests | |
2688 | * @inode: inode the direct I/O happens on | |
2689 | * | |
2690 | * This is called once we've finished processing a direct I/O request, | |
2691 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
2692 | */ | |
2693 | static inline void inode_dio_end(struct inode *inode) | |
2694 | { | |
2695 | if (atomic_dec_and_test(&inode->i_dio_count)) | |
2696 | wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); | |
2697 | } | |
1d59d61f | 2698 | |
5f16f322 TT |
2699 | extern void inode_set_flags(struct inode *inode, unsigned int flags, |
2700 | unsigned int mask); | |
2701 | ||
4b6f5d20 | 2702 | extern const struct file_operations generic_ro_fops; |
1da177e4 LT |
2703 | |
2704 | #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) | |
2705 | ||
5d826c84 | 2706 | extern int readlink_copy(char __user *, int, const char *); |
1da177e4 | 2707 | extern int page_readlink(struct dentry *, char __user *, int); |
6e77137b | 2708 | extern const char *page_follow_link_light(struct dentry *, void **); |
5f2c4179 | 2709 | extern void page_put_link(struct inode *, void *); |
0adb25d2 | 2710 | extern int __page_symlink(struct inode *inode, const char *symname, int len, |
54566b2c | 2711 | int nofs); |
1da177e4 | 2712 | extern int page_symlink(struct inode *inode, const char *symname, int len); |
c5ef1c42 | 2713 | extern const struct inode_operations page_symlink_inode_operations; |
5f2c4179 | 2714 | extern void kfree_put_link(struct inode *, void *); |
ecc087ff | 2715 | extern void free_page_put_link(struct inode *, void *); |
1da177e4 LT |
2716 | extern int generic_readlink(struct dentry *, char __user *, int); |
2717 | extern void generic_fillattr(struct inode *, struct kstat *); | |
b7a6ec52 | 2718 | int vfs_getattr_nosec(struct path *path, struct kstat *stat); |
3dadecce | 2719 | extern int vfs_getattr(struct path *, struct kstat *); |
b462707e | 2720 | void __inode_add_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 2721 | void inode_add_bytes(struct inode *inode, loff_t bytes); |
1c8924eb | 2722 | void __inode_sub_bytes(struct inode *inode, loff_t bytes); |
1da177e4 LT |
2723 | void inode_sub_bytes(struct inode *inode, loff_t bytes); |
2724 | loff_t inode_get_bytes(struct inode *inode); | |
2725 | void inode_set_bytes(struct inode *inode, loff_t bytes); | |
6e77137b | 2726 | const char *simple_follow_link(struct dentry *, void **); |
61ba64fc | 2727 | extern const struct inode_operations simple_symlink_inode_operations; |
1da177e4 | 2728 | |
5c0ba4e0 | 2729 | extern int iterate_dir(struct file *, struct dir_context *); |
1da177e4 | 2730 | |
c7887325 DH |
2731 | extern int vfs_stat(const char __user *, struct kstat *); |
2732 | extern int vfs_lstat(const char __user *, struct kstat *); | |
1da177e4 | 2733 | extern int vfs_fstat(unsigned int, struct kstat *); |
c7887325 | 2734 | extern int vfs_fstatat(int , const char __user *, struct kstat *, int); |
1da177e4 | 2735 | |
deb21db7 EZ |
2736 | extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd, |
2737 | unsigned long arg); | |
e9079cce | 2738 | extern int __generic_block_fiemap(struct inode *inode, |
3a3076f4 JB |
2739 | struct fiemap_extent_info *fieinfo, |
2740 | loff_t start, loff_t len, | |
2741 | get_block_t *get_block); | |
68c9d702 JB |
2742 | extern int generic_block_fiemap(struct inode *inode, |
2743 | struct fiemap_extent_info *fieinfo, u64 start, | |
2744 | u64 len, get_block_t *get_block); | |
1da177e4 | 2745 | |
c18479fe MS |
2746 | extern void get_filesystem(struct file_system_type *fs); |
2747 | extern void put_filesystem(struct file_system_type *fs); | |
1da177e4 LT |
2748 | extern struct file_system_type *get_fs_type(const char *name); |
2749 | extern struct super_block *get_super(struct block_device *); | |
6b6dc836 | 2750 | extern struct super_block *get_super_thawed(struct block_device *); |
4504230a | 2751 | extern struct super_block *get_active_super(struct block_device *bdev); |
1da177e4 | 2752 | extern void drop_super(struct super_block *sb); |
01a05b33 | 2753 | extern void iterate_supers(void (*)(struct super_block *, void *), void *); |
43e15cdb AV |
2754 | extern void iterate_supers_type(struct file_system_type *, |
2755 | void (*)(struct super_block *, void *), void *); | |
1da177e4 LT |
2756 | |
2757 | extern int dcache_dir_open(struct inode *, struct file *); | |
2758 | extern int dcache_dir_close(struct inode *, struct file *); | |
2759 | extern loff_t dcache_dir_lseek(struct file *, loff_t, int); | |
5f99f4e7 | 2760 | extern int dcache_readdir(struct file *, struct dir_context *); |
7bb46a67 | 2761 | extern int simple_setattr(struct dentry *, struct iattr *); |
1da177e4 | 2762 | extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *); |
726c3342 | 2763 | extern int simple_statfs(struct dentry *, struct kstatfs *); |
20955e89 | 2764 | extern int simple_open(struct inode *inode, struct file *file); |
1da177e4 LT |
2765 | extern int simple_link(struct dentry *, struct inode *, struct dentry *); |
2766 | extern int simple_unlink(struct inode *, struct dentry *); | |
2767 | extern int simple_rmdir(struct inode *, struct dentry *); | |
2768 | extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); | |
02c24a82 | 2769 | extern int noop_fsync(struct file *, loff_t, loff_t, int); |
1da177e4 LT |
2770 | extern int simple_empty(struct dentry *); |
2771 | extern int simple_readpage(struct file *file, struct page *page); | |
afddba49 NP |
2772 | extern int simple_write_begin(struct file *file, struct address_space *mapping, |
2773 | loff_t pos, unsigned len, unsigned flags, | |
2774 | struct page **pagep, void **fsdata); | |
2775 | extern int simple_write_end(struct file *file, struct address_space *mapping, | |
2776 | loff_t pos, unsigned len, unsigned copied, | |
2777 | struct page *page, void *fsdata); | |
b26d4cd3 | 2778 | extern int always_delete_dentry(const struct dentry *); |
6987843f | 2779 | extern struct inode *alloc_anon_inode(struct super_block *); |
e6f5c789 | 2780 | extern int simple_nosetlease(struct file *, long, struct file_lock **, void **); |
b26d4cd3 | 2781 | extern const struct dentry_operations simple_dentry_operations; |
1da177e4 | 2782 | |
00cd8dd3 | 2783 | extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); |
1da177e4 | 2784 | extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); |
4b6f5d20 | 2785 | extern const struct file_operations simple_dir_operations; |
c5ef1c42 | 2786 | extern const struct inode_operations simple_dir_inode_operations; |
4b6f5d20 | 2787 | struct tree_descr { char *name; const struct file_operations *ops; int mode; }; |
1da177e4 | 2788 | struct dentry *d_alloc_name(struct dentry *, const char *); |
7d683a09 | 2789 | extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *); |
1f5ce9e9 | 2790 | extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); |
1da177e4 LT |
2791 | extern void simple_release_fs(struct vfsmount **mount, int *count); |
2792 | ||
93b07113 AM |
2793 | extern ssize_t simple_read_from_buffer(void __user *to, size_t count, |
2794 | loff_t *ppos, const void *from, size_t available); | |
6a727b43 JS |
2795 | extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, |
2796 | const void __user *from, size_t count); | |
1da177e4 | 2797 | |
ac13a829 | 2798 | extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); |
02c24a82 | 2799 | extern int generic_file_fsync(struct file *, loff_t, loff_t, int); |
d5aacad5 | 2800 | |
30ca22c7 PL |
2801 | extern int generic_check_addressable(unsigned, u64); |
2802 | ||
e965f963 | 2803 | #ifdef CONFIG_MIGRATION |
2d1db3b1 | 2804 | extern int buffer_migrate_page(struct address_space *, |
a6bc32b8 MG |
2805 | struct page *, struct page *, |
2806 | enum migrate_mode); | |
e965f963 CL |
2807 | #else |
2808 | #define buffer_migrate_page NULL | |
2809 | #endif | |
2810 | ||
25d9e2d1 | 2811 | extern int inode_change_ok(const struct inode *, struct iattr *); |
2812 | extern int inode_newsize_ok(const struct inode *, loff_t offset); | |
6a1a90ad | 2813 | extern void setattr_copy(struct inode *inode, const struct iattr *attr); |
1da177e4 | 2814 | |
c3b2da31 | 2815 | extern int file_update_time(struct file *file); |
1da177e4 | 2816 | |
34c80b1d | 2817 | extern int generic_show_options(struct seq_file *m, struct dentry *root); |
b3b304a2 | 2818 | extern void save_mount_options(struct super_block *sb, char *options); |
2a32cebd | 2819 | extern void replace_mount_options(struct super_block *sb, char *options); |
b3b304a2 | 2820 | |
d475c634 MW |
2821 | static inline bool io_is_direct(struct file *filp) |
2822 | { | |
2823 | return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp)); | |
2824 | } | |
2825 | ||
2ba48ce5 AV |
2826 | static inline int iocb_flags(struct file *file) |
2827 | { | |
2828 | int res = 0; | |
2829 | if (file->f_flags & O_APPEND) | |
2830 | res |= IOCB_APPEND; | |
2831 | if (io_is_direct(file)) | |
2832 | res |= IOCB_DIRECT; | |
2833 | return res; | |
2834 | } | |
2835 | ||
1da177e4 LT |
2836 | static inline ino_t parent_ino(struct dentry *dentry) |
2837 | { | |
2838 | ino_t res; | |
2839 | ||
b5c84bf6 NP |
2840 | /* |
2841 | * Don't strictly need d_lock here? If the parent ino could change | |
2842 | * then surely we'd have a deeper race in the caller? | |
2843 | */ | |
1da177e4 LT |
2844 | spin_lock(&dentry->d_lock); |
2845 | res = dentry->d_parent->d_inode->i_ino; | |
2846 | spin_unlock(&dentry->d_lock); | |
2847 | return res; | |
2848 | } | |
2849 | ||
1da177e4 LT |
2850 | /* Transaction based IO helpers */ |
2851 | ||
2852 | /* | |
2853 | * An argresp is stored in an allocated page and holds the | |
2854 | * size of the argument or response, along with its content | |
2855 | */ | |
2856 | struct simple_transaction_argresp { | |
2857 | ssize_t size; | |
2858 | char data[0]; | |
2859 | }; | |
2860 | ||
2861 | #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) | |
2862 | ||
2863 | char *simple_transaction_get(struct file *file, const char __user *buf, | |
2864 | size_t size); | |
2865 | ssize_t simple_transaction_read(struct file *file, char __user *buf, | |
2866 | size_t size, loff_t *pos); | |
2867 | int simple_transaction_release(struct inode *inode, struct file *file); | |
2868 | ||
76791ab2 | 2869 | void simple_transaction_set(struct file *file, size_t n); |
1da177e4 | 2870 | |
acaefc25 AB |
2871 | /* |
2872 | * simple attribute files | |
2873 | * | |
2874 | * These attributes behave similar to those in sysfs: | |
2875 | * | |
2876 | * Writing to an attribute immediately sets a value, an open file can be | |
2877 | * written to multiple times. | |
2878 | * | |
2879 | * Reading from an attribute creates a buffer from the value that might get | |
2880 | * read with multiple read calls. When the attribute has been read | |
2881 | * completely, no further read calls are possible until the file is opened | |
2882 | * again. | |
2883 | * | |
2884 | * All attributes contain a text representation of a numeric value | |
2885 | * that are accessed with the get() and set() functions. | |
2886 | */ | |
2887 | #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ | |
2888 | static int __fops ## _open(struct inode *inode, struct file *file) \ | |
2889 | { \ | |
2890 | __simple_attr_check_format(__fmt, 0ull); \ | |
2891 | return simple_attr_open(inode, file, __get, __set, __fmt); \ | |
2892 | } \ | |
828c0950 | 2893 | static const struct file_operations __fops = { \ |
acaefc25 AB |
2894 | .owner = THIS_MODULE, \ |
2895 | .open = __fops ## _open, \ | |
74bedc4d | 2896 | .release = simple_attr_release, \ |
acaefc25 AB |
2897 | .read = simple_attr_read, \ |
2898 | .write = simple_attr_write, \ | |
1ec5584e | 2899 | .llseek = generic_file_llseek, \ |
68be3029 | 2900 | } |
acaefc25 | 2901 | |
b9075fa9 JP |
2902 | static inline __printf(1, 2) |
2903 | void __simple_attr_check_format(const char *fmt, ...) | |
acaefc25 AB |
2904 | { |
2905 | /* don't do anything, just let the compiler check the arguments; */ | |
2906 | } | |
2907 | ||
2908 | int simple_attr_open(struct inode *inode, struct file *file, | |
8b88b099 | 2909 | int (*get)(void *, u64 *), int (*set)(void *, u64), |
acaefc25 | 2910 | const char *fmt); |
74bedc4d | 2911 | int simple_attr_release(struct inode *inode, struct file *file); |
acaefc25 AB |
2912 | ssize_t simple_attr_read(struct file *file, char __user *buf, |
2913 | size_t len, loff_t *ppos); | |
2914 | ssize_t simple_attr_write(struct file *file, const char __user *buf, | |
2915 | size_t len, loff_t *ppos); | |
2916 | ||
4be28540 | 2917 | struct ctl_table; |
8d65af78 | 2918 | int proc_nr_files(struct ctl_table *table, int write, |
62239ac2 | 2919 | void __user *buffer, size_t *lenp, loff_t *ppos); |
312d3ca8 CH |
2920 | int proc_nr_dentry(struct ctl_table *table, int write, |
2921 | void __user *buffer, size_t *lenp, loff_t *ppos); | |
cffbc8aa DC |
2922 | int proc_nr_inodes(struct ctl_table *table, int write, |
2923 | void __user *buffer, size_t *lenp, loff_t *ppos); | |
38e23c95 | 2924 | int __init get_filesystem_list(char *buf); |
62239ac2 | 2925 | |
3cd90ea4 | 2926 | #define __FMODE_EXEC ((__force int) FMODE_EXEC) |
1a44bc8c NK |
2927 | #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) |
2928 | ||
6d125529 | 2929 | #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) |
ecf081d1 | 2930 | #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ |
1a44bc8c | 2931 | (flag & __FMODE_NONOTIFY))) |
5300990c | 2932 | |
8d334acd | 2933 | static inline int is_sxid(umode_t mode) |
69b45732 AK |
2934 | { |
2935 | return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP)); | |
2936 | } | |
2937 | ||
cbdf35bc MS |
2938 | static inline int check_sticky(struct inode *dir, struct inode *inode) |
2939 | { | |
2940 | if (!(dir->i_mode & S_ISVTX)) | |
2941 | return 0; | |
2942 | ||
2943 | return __check_sticky(dir, inode); | |
2944 | } | |
2945 | ||
69b45732 AK |
2946 | static inline void inode_has_no_xattr(struct inode *inode) |
2947 | { | |
9e1f1de0 | 2948 | if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC)) |
69b45732 AK |
2949 | inode->i_flags |= S_NOSEC; |
2950 | } | |
2951 | ||
a7400222 AV |
2952 | static inline bool is_root_inode(struct inode *inode) |
2953 | { | |
2954 | return inode == inode->i_sb->s_root->d_inode; | |
2955 | } | |
2956 | ||
5f99f4e7 AV |
2957 | static inline bool dir_emit(struct dir_context *ctx, |
2958 | const char *name, int namelen, | |
2959 | u64 ino, unsigned type) | |
2960 | { | |
2961 | return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0; | |
2962 | } | |
2963 | static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) | |
2964 | { | |
2965 | return ctx->actor(ctx, ".", 1, ctx->pos, | |
2966 | file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0; | |
2967 | } | |
2968 | static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) | |
2969 | { | |
2970 | return ctx->actor(ctx, "..", 2, ctx->pos, | |
2971 | parent_ino(file->f_path.dentry), DT_DIR) == 0; | |
2972 | } | |
2973 | static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) | |
2974 | { | |
2975 | if (ctx->pos == 0) { | |
2976 | if (!dir_emit_dot(file, ctx)) | |
2977 | return false; | |
2978 | ctx->pos = 1; | |
2979 | } | |
2980 | if (ctx->pos == 1) { | |
2981 | if (!dir_emit_dotdot(file, ctx)) | |
2982 | return false; | |
2983 | ctx->pos = 2; | |
2984 | } | |
2985 | return true; | |
2986 | } | |
5ded75ec AV |
2987 | static inline bool dir_relax(struct inode *inode) |
2988 | { | |
2989 | mutex_unlock(&inode->i_mutex); | |
2990 | mutex_lock(&inode->i_mutex); | |
2991 | return !IS_DEADDIR(inode); | |
2992 | } | |
5f99f4e7 | 2993 | |
1da177e4 | 2994 | #endif /* _LINUX_FS_H */ |