]>
Commit | Line | Data |
---|---|---|
1 | #ifndef _LINUX_FS_H | |
2 | #define _LINUX_FS_H | |
3 | ||
4 | ||
5 | #include <linux/linkage.h> | |
6 | #include <linux/wait.h> | |
7 | #include <linux/kdev_t.h> | |
8 | #include <linux/dcache.h> | |
9 | #include <linux/path.h> | |
10 | #include <linux/stat.h> | |
11 | #include <linux/cache.h> | |
12 | #include <linux/list.h> | |
13 | #include <linux/list_lru.h> | |
14 | #include <linux/llist.h> | |
15 | #include <linux/radix-tree.h> | |
16 | #include <linux/rbtree.h> | |
17 | #include <linux/init.h> | |
18 | #include <linux/pid.h> | |
19 | #include <linux/bug.h> | |
20 | #include <linux/mutex.h> | |
21 | #include <linux/rwsem.h> | |
22 | #include <linux/capability.h> | |
23 | #include <linux/semaphore.h> | |
24 | #include <linux/fiemap.h> | |
25 | #include <linux/rculist_bl.h> | |
26 | #include <linux/atomic.h> | |
27 | #include <linux/shrinker.h> | |
28 | #include <linux/migrate_mode.h> | |
29 | #include <linux/uidgid.h> | |
30 | #include <linux/lockdep.h> | |
31 | #include <linux/percpu-rwsem.h> | |
32 | #include <linux/blk_types.h> | |
33 | ||
34 | #include <asm/byteorder.h> | |
35 | #include <uapi/linux/fs.h> | |
36 | ||
37 | struct backing_dev_info; | |
38 | struct export_operations; | |
39 | struct hd_geometry; | |
40 | struct iovec; | |
41 | struct nameidata; | |
42 | struct kiocb; | |
43 | struct kobject; | |
44 | struct pipe_inode_info; | |
45 | struct poll_table_struct; | |
46 | struct kstatfs; | |
47 | struct vm_area_struct; | |
48 | struct vfsmount; | |
49 | struct cred; | |
50 | struct swap_info_struct; | |
51 | struct seq_file; | |
52 | struct workqueue_struct; | |
53 | struct iov_iter; | |
54 | ||
55 | extern void __init inode_init(void); | |
56 | extern void __init inode_init_early(void); | |
57 | extern void __init files_init(unsigned long); | |
58 | ||
59 | extern struct files_stat_struct files_stat; | |
60 | extern unsigned long get_max_files(void); | |
61 | extern int sysctl_nr_open; | |
62 | extern struct inodes_stat_t inodes_stat; | |
63 | extern int leases_enable, lease_break_time; | |
64 | extern int sysctl_protected_symlinks; | |
65 | extern int sysctl_protected_hardlinks; | |
66 | ||
67 | struct buffer_head; | |
68 | typedef int (get_block_t)(struct inode *inode, sector_t iblock, | |
69 | struct buffer_head *bh_result, int create); | |
70 | typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset, | |
71 | ssize_t bytes, void *private); | |
72 | ||
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 | ||
129 | /* File needs atomic accesses to f_pos */ | |
130 | #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) | |
131 | /* Write access to underlying fs */ | |
132 | #define FMODE_WRITER ((__force fmode_t)0x10000) | |
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) | |
137 | ||
138 | /* File was opened by fanotify and shouldn't generate fanotify events */ | |
139 | #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) | |
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 | ||
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 | |
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 | ||
205 | /* | |
206 | * Attribute flags. These should be or-ed together to figure out what | |
207 | * has been changed! | |
208 | */ | |
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) | |
226 | ||
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 | ||
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; | |
246 | kuid_t ia_uid; | |
247 | kgid_t ia_gid; | |
248 | loff_t ia_size; | |
249 | struct timespec ia_atime; | |
250 | struct timespec ia_mtime; | |
251 | struct timespec ia_ctime; | |
252 | ||
253 | /* | |
254 | * Not an attribute, but an auxiliary info for filesystems wanting to | |
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; | |
259 | }; | |
260 | ||
261 | /* | |
262 | * Includes for diskquotas. | |
263 | */ | |
264 | #include <linux/quota.h> | |
265 | ||
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 | ||
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 | |
290 | * by readpage(). | |
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 | ||
303 | #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */ | |
304 | #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */ | |
305 | #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct | |
306 | * helper code (eg buffer layer) | |
307 | * to clear GFP_FS from alloc */ | |
308 | ||
309 | /* | |
310 | * oh the beauties of C type declarations. | |
311 | */ | |
312 | struct page; | |
313 | struct address_space; | |
314 | struct writeback_control; | |
315 | ||
316 | /* | |
317 | * "descriptor" for what we're up to with a read. | |
318 | * This allows us to use the same read code yet | |
319 | * have multiple different users of the data that | |
320 | * we read from a file. | |
321 | * | |
322 | * The simplest case just copies the data to user | |
323 | * mode. | |
324 | */ | |
325 | typedef struct { | |
326 | size_t written; | |
327 | size_t count; | |
328 | union { | |
329 | char __user *buf; | |
330 | void *data; | |
331 | } arg; | |
332 | int error; | |
333 | } read_descriptor_t; | |
334 | ||
335 | typedef int (*read_actor_t)(read_descriptor_t *, struct page *, | |
336 | unsigned long, unsigned long); | |
337 | ||
338 | struct address_space_operations { | |
339 | int (*writepage)(struct page *page, struct writeback_control *wbc); | |
340 | int (*readpage)(struct file *, struct page *); | |
341 | ||
342 | /* Write back some dirty pages from this mapping. */ | |
343 | int (*writepages)(struct address_space *, struct writeback_control *); | |
344 | ||
345 | /* Set a page dirty. Return true if this dirtied it */ | |
346 | int (*set_page_dirty)(struct page *page); | |
347 | ||
348 | int (*readpages)(struct file *filp, struct address_space *mapping, | |
349 | struct list_head *pages, unsigned nr_pages); | |
350 | ||
351 | int (*write_begin)(struct file *, struct address_space *mapping, | |
352 | loff_t pos, unsigned len, unsigned flags, | |
353 | struct page **pagep, void **fsdata); | |
354 | int (*write_end)(struct file *, struct address_space *mapping, | |
355 | loff_t pos, unsigned len, unsigned copied, | |
356 | struct page *page, void *fsdata); | |
357 | ||
358 | /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ | |
359 | sector_t (*bmap)(struct address_space *, sector_t); | |
360 | void (*invalidatepage) (struct page *, unsigned int, unsigned int); | |
361 | int (*releasepage) (struct page *, gfp_t); | |
362 | void (*freepage)(struct page *); | |
363 | ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset); | |
364 | int (*get_xip_mem)(struct address_space *, pgoff_t, int, | |
365 | void **, unsigned long *); | |
366 | /* | |
367 | * migrate the contents of a page to the specified target. If | |
368 | * migrate_mode is MIGRATE_ASYNC, it must not block. | |
369 | */ | |
370 | int (*migratepage) (struct address_space *, | |
371 | struct page *, struct page *, enum migrate_mode); | |
372 | int (*launder_page) (struct page *); | |
373 | int (*is_partially_uptodate) (struct page *, unsigned long, | |
374 | unsigned long); | |
375 | void (*is_dirty_writeback) (struct page *, bool *, bool *); | |
376 | int (*error_remove_page)(struct address_space *, struct page *); | |
377 | ||
378 | /* swapfile support */ | |
379 | int (*swap_activate)(struct swap_info_struct *sis, struct file *file, | |
380 | sector_t *span); | |
381 | void (*swap_deactivate)(struct file *file); | |
382 | }; | |
383 | ||
384 | extern const struct address_space_operations empty_aops; | |
385 | ||
386 | /* | |
387 | * pagecache_write_begin/pagecache_write_end must be used by general code | |
388 | * to write into the pagecache. | |
389 | */ | |
390 | int pagecache_write_begin(struct file *, struct address_space *mapping, | |
391 | loff_t pos, unsigned len, unsigned flags, | |
392 | struct page **pagep, void **fsdata); | |
393 | ||
394 | int pagecache_write_end(struct file *, struct address_space *mapping, | |
395 | loff_t pos, unsigned len, unsigned copied, | |
396 | struct page *page, void *fsdata); | |
397 | ||
398 | struct address_space { | |
399 | struct inode *host; /* owner: inode, block_device */ | |
400 | struct radix_tree_root page_tree; /* radix tree of all pages */ | |
401 | spinlock_t tree_lock; /* and lock protecting it */ | |
402 | atomic_t i_mmap_writable;/* count VM_SHARED mappings */ | |
403 | struct rb_root i_mmap; /* tree of private and shared mappings */ | |
404 | struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */ | |
405 | struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */ | |
406 | /* Protected by tree_lock together with the radix tree */ | |
407 | unsigned long nrpages; /* number of total pages */ | |
408 | unsigned long nrshadows; /* number of shadow entries */ | |
409 | pgoff_t writeback_index;/* writeback starts here */ | |
410 | const struct address_space_operations *a_ops; /* methods */ | |
411 | unsigned long flags; /* error bits/gfp mask */ | |
412 | spinlock_t private_lock; /* for use by the address_space */ | |
413 | struct list_head private_list; /* ditto */ | |
414 | void *private_data; /* ditto */ | |
415 | } __attribute__((aligned(sizeof(long)))); | |
416 | /* | |
417 | * On most architectures that alignment is already the case; but | |
418 | * must be enforced here for CRIS, to let the least significant bit | |
419 | * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. | |
420 | */ | |
421 | struct request_queue; | |
422 | ||
423 | struct block_device { | |
424 | dev_t bd_dev; /* not a kdev_t - it's a search key */ | |
425 | int bd_openers; | |
426 | struct inode * bd_inode; /* will die */ | |
427 | struct super_block * bd_super; | |
428 | struct mutex bd_mutex; /* open/close mutex */ | |
429 | struct list_head bd_inodes; | |
430 | void * bd_claiming; | |
431 | void * bd_holder; | |
432 | int bd_holders; | |
433 | bool bd_write_holder; | |
434 | #ifdef CONFIG_SYSFS | |
435 | struct list_head bd_holder_disks; | |
436 | #endif | |
437 | struct block_device * bd_contains; | |
438 | unsigned bd_block_size; | |
439 | struct hd_struct * bd_part; | |
440 | /* number of times partitions within this device have been opened. */ | |
441 | unsigned bd_part_count; | |
442 | int bd_invalidated; | |
443 | struct gendisk * bd_disk; | |
444 | struct request_queue * bd_queue; | |
445 | struct list_head bd_list; | |
446 | /* | |
447 | * Private data. You must have bd_claim'ed the block_device | |
448 | * to use this. NOTE: bd_claim allows an owner to claim | |
449 | * the same device multiple times, the owner must take special | |
450 | * care to not mess up bd_private for that case. | |
451 | */ | |
452 | unsigned long bd_private; | |
453 | ||
454 | /* The counter of freeze processes */ | |
455 | int bd_fsfreeze_count; | |
456 | /* Mutex for freeze */ | |
457 | struct mutex bd_fsfreeze_mutex; | |
458 | }; | |
459 | ||
460 | /* | |
461 | * Radix-tree tags, for tagging dirty and writeback pages within the pagecache | |
462 | * radix trees | |
463 | */ | |
464 | #define PAGECACHE_TAG_DIRTY 0 | |
465 | #define PAGECACHE_TAG_WRITEBACK 1 | |
466 | #define PAGECACHE_TAG_TOWRITE 2 | |
467 | ||
468 | int mapping_tagged(struct address_space *mapping, int tag); | |
469 | ||
470 | static inline void i_mmap_lock_write(struct address_space *mapping) | |
471 | { | |
472 | down_write(&mapping->i_mmap_rwsem); | |
473 | } | |
474 | ||
475 | static inline void i_mmap_unlock_write(struct address_space *mapping) | |
476 | { | |
477 | up_write(&mapping->i_mmap_rwsem); | |
478 | } | |
479 | ||
480 | static inline void i_mmap_lock_read(struct address_space *mapping) | |
481 | { | |
482 | down_read(&mapping->i_mmap_rwsem); | |
483 | } | |
484 | ||
485 | static inline void i_mmap_unlock_read(struct address_space *mapping) | |
486 | { | |
487 | up_read(&mapping->i_mmap_rwsem); | |
488 | } | |
489 | ||
490 | /* | |
491 | * Might pages of this file be mapped into userspace? | |
492 | */ | |
493 | static inline int mapping_mapped(struct address_space *mapping) | |
494 | { | |
495 | return !RB_EMPTY_ROOT(&mapping->i_mmap) || | |
496 | !list_empty(&mapping->i_mmap_nonlinear); | |
497 | } | |
498 | ||
499 | /* | |
500 | * Might pages of this file have been modified in userspace? | |
501 | * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff | |
502 | * marks vma as VM_SHARED if it is shared, and the file was opened for | |
503 | * writing i.e. vma may be mprotected writable even if now readonly. | |
504 | * | |
505 | * If i_mmap_writable is negative, no new writable mappings are allowed. You | |
506 | * can only deny writable mappings, if none exists right now. | |
507 | */ | |
508 | static inline int mapping_writably_mapped(struct address_space *mapping) | |
509 | { | |
510 | return atomic_read(&mapping->i_mmap_writable) > 0; | |
511 | } | |
512 | ||
513 | static inline int mapping_map_writable(struct address_space *mapping) | |
514 | { | |
515 | return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? | |
516 | 0 : -EPERM; | |
517 | } | |
518 | ||
519 | static inline void mapping_unmap_writable(struct address_space *mapping) | |
520 | { | |
521 | atomic_dec(&mapping->i_mmap_writable); | |
522 | } | |
523 | ||
524 | static inline int mapping_deny_writable(struct address_space *mapping) | |
525 | { | |
526 | return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? | |
527 | 0 : -EBUSY; | |
528 | } | |
529 | ||
530 | static inline void mapping_allow_writable(struct address_space *mapping) | |
531 | { | |
532 | atomic_inc(&mapping->i_mmap_writable); | |
533 | } | |
534 | ||
535 | /* | |
536 | * Use sequence counter to get consistent i_size on 32-bit processors. | |
537 | */ | |
538 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
539 | #include <linux/seqlock.h> | |
540 | #define __NEED_I_SIZE_ORDERED | |
541 | #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) | |
542 | #else | |
543 | #define i_size_ordered_init(inode) do { } while (0) | |
544 | #endif | |
545 | ||
546 | struct posix_acl; | |
547 | #define ACL_NOT_CACHED ((void *)(-1)) | |
548 | ||
549 | #define IOP_FASTPERM 0x0001 | |
550 | #define IOP_LOOKUP 0x0002 | |
551 | #define IOP_NOFOLLOW 0x0004 | |
552 | ||
553 | /* | |
554 | * Keep mostly read-only and often accessed (especially for | |
555 | * the RCU path lookup and 'stat' data) fields at the beginning | |
556 | * of the 'struct inode' | |
557 | */ | |
558 | struct inode { | |
559 | umode_t i_mode; | |
560 | unsigned short i_opflags; | |
561 | kuid_t i_uid; | |
562 | kgid_t i_gid; | |
563 | unsigned int i_flags; | |
564 | ||
565 | #ifdef CONFIG_FS_POSIX_ACL | |
566 | struct posix_acl *i_acl; | |
567 | struct posix_acl *i_default_acl; | |
568 | #endif | |
569 | ||
570 | const struct inode_operations *i_op; | |
571 | struct super_block *i_sb; | |
572 | struct address_space *i_mapping; | |
573 | ||
574 | #ifdef CONFIG_SECURITY | |
575 | void *i_security; | |
576 | #endif | |
577 | ||
578 | /* Stat data, not accessed from path walking */ | |
579 | unsigned long i_ino; | |
580 | /* | |
581 | * Filesystems may only read i_nlink directly. They shall use the | |
582 | * following functions for modification: | |
583 | * | |
584 | * (set|clear|inc|drop)_nlink | |
585 | * inode_(inc|dec)_link_count | |
586 | */ | |
587 | union { | |
588 | const unsigned int i_nlink; | |
589 | unsigned int __i_nlink; | |
590 | }; | |
591 | dev_t i_rdev; | |
592 | loff_t i_size; | |
593 | struct timespec i_atime; | |
594 | struct timespec i_mtime; | |
595 | struct timespec i_ctime; | |
596 | spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ | |
597 | unsigned short i_bytes; | |
598 | unsigned int i_blkbits; | |
599 | blkcnt_t i_blocks; | |
600 | ||
601 | #ifdef __NEED_I_SIZE_ORDERED | |
602 | seqcount_t i_size_seqcount; | |
603 | #endif | |
604 | ||
605 | /* Misc */ | |
606 | unsigned long i_state; | |
607 | struct mutex i_mutex; | |
608 | ||
609 | unsigned long dirtied_when; /* jiffies of first dirtying */ | |
610 | ||
611 | struct hlist_node i_hash; | |
612 | struct list_head i_wb_list; /* backing dev IO list */ | |
613 | struct list_head i_lru; /* inode LRU list */ | |
614 | struct list_head i_sb_list; | |
615 | union { | |
616 | struct hlist_head i_dentry; | |
617 | struct rcu_head i_rcu; | |
618 | }; | |
619 | u64 i_version; | |
620 | atomic_t i_count; | |
621 | atomic_t i_dio_count; | |
622 | atomic_t i_writecount; | |
623 | #ifdef CONFIG_IMA | |
624 | atomic_t i_readcount; /* struct files open RO */ | |
625 | #endif | |
626 | const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ | |
627 | struct file_lock *i_flock; | |
628 | struct address_space i_data; | |
629 | struct list_head i_devices; | |
630 | union { | |
631 | struct pipe_inode_info *i_pipe; | |
632 | struct block_device *i_bdev; | |
633 | struct cdev *i_cdev; | |
634 | }; | |
635 | ||
636 | __u32 i_generation; | |
637 | ||
638 | #ifdef CONFIG_FSNOTIFY | |
639 | __u32 i_fsnotify_mask; /* all events this inode cares about */ | |
640 | struct hlist_head i_fsnotify_marks; | |
641 | #endif | |
642 | ||
643 | void *i_private; /* fs or device private pointer */ | |
644 | }; | |
645 | ||
646 | static inline int inode_unhashed(struct inode *inode) | |
647 | { | |
648 | return hlist_unhashed(&inode->i_hash); | |
649 | } | |
650 | ||
651 | /* | |
652 | * inode->i_mutex nesting subclasses for the lock validator: | |
653 | * | |
654 | * 0: the object of the current VFS operation | |
655 | * 1: parent | |
656 | * 2: child/target | |
657 | * 3: xattr | |
658 | * 4: second non-directory | |
659 | * 5: second parent (when locking independent directories in rename) | |
660 | * | |
661 | * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two | |
662 | * non-directories at once. | |
663 | * | |
664 | * The locking order between these classes is | |
665 | * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory | |
666 | */ | |
667 | enum inode_i_mutex_lock_class | |
668 | { | |
669 | I_MUTEX_NORMAL, | |
670 | I_MUTEX_PARENT, | |
671 | I_MUTEX_CHILD, | |
672 | I_MUTEX_XATTR, | |
673 | I_MUTEX_NONDIR2, | |
674 | I_MUTEX_PARENT2, | |
675 | }; | |
676 | ||
677 | void lock_two_nondirectories(struct inode *, struct inode*); | |
678 | void unlock_two_nondirectories(struct inode *, struct inode*); | |
679 | ||
680 | /* | |
681 | * NOTE: in a 32bit arch with a preemptable kernel and | |
682 | * an UP compile the i_size_read/write must be atomic | |
683 | * with respect to the local cpu (unlike with preempt disabled), | |
684 | * but they don't need to be atomic with respect to other cpus like in | |
685 | * true SMP (so they need either to either locally disable irq around | |
686 | * the read or for example on x86 they can be still implemented as a | |
687 | * cmpxchg8b without the need of the lock prefix). For SMP compiles | |
688 | * and 64bit archs it makes no difference if preempt is enabled or not. | |
689 | */ | |
690 | static inline loff_t i_size_read(const struct inode *inode) | |
691 | { | |
692 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
693 | loff_t i_size; | |
694 | unsigned int seq; | |
695 | ||
696 | do { | |
697 | seq = read_seqcount_begin(&inode->i_size_seqcount); | |
698 | i_size = inode->i_size; | |
699 | } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); | |
700 | return i_size; | |
701 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) | |
702 | loff_t i_size; | |
703 | ||
704 | preempt_disable(); | |
705 | i_size = inode->i_size; | |
706 | preempt_enable(); | |
707 | return i_size; | |
708 | #else | |
709 | return inode->i_size; | |
710 | #endif | |
711 | } | |
712 | ||
713 | /* | |
714 | * NOTE: unlike i_size_read(), i_size_write() does need locking around it | |
715 | * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount | |
716 | * can be lost, resulting in subsequent i_size_read() calls spinning forever. | |
717 | */ | |
718 | static inline void i_size_write(struct inode *inode, loff_t i_size) | |
719 | { | |
720 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
721 | preempt_disable(); | |
722 | write_seqcount_begin(&inode->i_size_seqcount); | |
723 | inode->i_size = i_size; | |
724 | write_seqcount_end(&inode->i_size_seqcount); | |
725 | preempt_enable(); | |
726 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) | |
727 | preempt_disable(); | |
728 | inode->i_size = i_size; | |
729 | preempt_enable(); | |
730 | #else | |
731 | inode->i_size = i_size; | |
732 | #endif | |
733 | } | |
734 | ||
735 | /* Helper functions so that in most cases filesystems will | |
736 | * not need to deal directly with kuid_t and kgid_t and can | |
737 | * instead deal with the raw numeric values that are stored | |
738 | * in the filesystem. | |
739 | */ | |
740 | static inline uid_t i_uid_read(const struct inode *inode) | |
741 | { | |
742 | return from_kuid(&init_user_ns, inode->i_uid); | |
743 | } | |
744 | ||
745 | static inline gid_t i_gid_read(const struct inode *inode) | |
746 | { | |
747 | return from_kgid(&init_user_ns, inode->i_gid); | |
748 | } | |
749 | ||
750 | static inline void i_uid_write(struct inode *inode, uid_t uid) | |
751 | { | |
752 | inode->i_uid = make_kuid(&init_user_ns, uid); | |
753 | } | |
754 | ||
755 | static inline void i_gid_write(struct inode *inode, gid_t gid) | |
756 | { | |
757 | inode->i_gid = make_kgid(&init_user_ns, gid); | |
758 | } | |
759 | ||
760 | static inline unsigned iminor(const struct inode *inode) | |
761 | { | |
762 | return MINOR(inode->i_rdev); | |
763 | } | |
764 | ||
765 | static inline unsigned imajor(const struct inode *inode) | |
766 | { | |
767 | return MAJOR(inode->i_rdev); | |
768 | } | |
769 | ||
770 | extern struct block_device *I_BDEV(struct inode *inode); | |
771 | ||
772 | struct fown_struct { | |
773 | rwlock_t lock; /* protects pid, uid, euid fields */ | |
774 | struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ | |
775 | enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ | |
776 | kuid_t uid, euid; /* uid/euid of process setting the owner */ | |
777 | int signum; /* posix.1b rt signal to be delivered on IO */ | |
778 | }; | |
779 | ||
780 | /* | |
781 | * Track a single file's readahead state | |
782 | */ | |
783 | struct file_ra_state { | |
784 | pgoff_t start; /* where readahead started */ | |
785 | unsigned int size; /* # of readahead pages */ | |
786 | unsigned int async_size; /* do asynchronous readahead when | |
787 | there are only # of pages ahead */ | |
788 | ||
789 | unsigned int ra_pages; /* Maximum readahead window */ | |
790 | unsigned int mmap_miss; /* Cache miss stat for mmap accesses */ | |
791 | loff_t prev_pos; /* Cache last read() position */ | |
792 | }; | |
793 | ||
794 | /* | |
795 | * Check if @index falls in the readahead windows. | |
796 | */ | |
797 | static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) | |
798 | { | |
799 | return (index >= ra->start && | |
800 | index < ra->start + ra->size); | |
801 | } | |
802 | ||
803 | struct file { | |
804 | union { | |
805 | struct llist_node fu_llist; | |
806 | struct rcu_head fu_rcuhead; | |
807 | } f_u; | |
808 | struct path f_path; | |
809 | struct inode *f_inode; /* cached value */ | |
810 | const struct file_operations *f_op; | |
811 | ||
812 | /* | |
813 | * Protects f_ep_links, f_flags. | |
814 | * Must not be taken from IRQ context. | |
815 | */ | |
816 | spinlock_t f_lock; | |
817 | atomic_long_t f_count; | |
818 | unsigned int f_flags; | |
819 | fmode_t f_mode; | |
820 | struct mutex f_pos_lock; | |
821 | loff_t f_pos; | |
822 | struct fown_struct f_owner; | |
823 | const struct cred *f_cred; | |
824 | struct file_ra_state f_ra; | |
825 | ||
826 | u64 f_version; | |
827 | #ifdef CONFIG_SECURITY | |
828 | void *f_security; | |
829 | #endif | |
830 | /* needed for tty driver, and maybe others */ | |
831 | void *private_data; | |
832 | ||
833 | #ifdef CONFIG_EPOLL | |
834 | /* Used by fs/eventpoll.c to link all the hooks to this file */ | |
835 | struct list_head f_ep_links; | |
836 | struct list_head f_tfile_llink; | |
837 | #endif /* #ifdef CONFIG_EPOLL */ | |
838 | struct address_space *f_mapping; | |
839 | } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ | |
840 | ||
841 | struct file_handle { | |
842 | __u32 handle_bytes; | |
843 | int handle_type; | |
844 | /* file identifier */ | |
845 | unsigned char f_handle[0]; | |
846 | }; | |
847 | ||
848 | static inline struct file *get_file(struct file *f) | |
849 | { | |
850 | atomic_long_inc(&f->f_count); | |
851 | return f; | |
852 | } | |
853 | #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1) | |
854 | #define file_count(x) atomic_long_read(&(x)->f_count) | |
855 | ||
856 | #define MAX_NON_LFS ((1UL<<31) - 1) | |
857 | ||
858 | /* Page cache limit. The filesystems should put that into their s_maxbytes | |
859 | limits, otherwise bad things can happen in VM. */ | |
860 | #if BITS_PER_LONG==32 | |
861 | #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1) | |
862 | #elif BITS_PER_LONG==64 | |
863 | #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL) | |
864 | #endif | |
865 | ||
866 | #define FL_POSIX 1 | |
867 | #define FL_FLOCK 2 | |
868 | #define FL_DELEG 4 /* NFSv4 delegation */ | |
869 | #define FL_ACCESS 8 /* not trying to lock, just looking */ | |
870 | #define FL_EXISTS 16 /* when unlocking, test for existence */ | |
871 | #define FL_LEASE 32 /* lease held on this file */ | |
872 | #define FL_CLOSE 64 /* unlock on close */ | |
873 | #define FL_SLEEP 128 /* A blocking lock */ | |
874 | #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */ | |
875 | #define FL_UNLOCK_PENDING 512 /* Lease is being broken */ | |
876 | #define FL_OFDLCK 1024 /* lock is "owned" by struct file */ | |
877 | ||
878 | /* | |
879 | * Special return value from posix_lock_file() and vfs_lock_file() for | |
880 | * asynchronous locking. | |
881 | */ | |
882 | #define FILE_LOCK_DEFERRED 1 | |
883 | ||
884 | /* legacy typedef, should eventually be removed */ | |
885 | typedef void *fl_owner_t; | |
886 | ||
887 | struct file_lock_operations { | |
888 | void (*fl_copy_lock)(struct file_lock *, struct file_lock *); | |
889 | void (*fl_release_private)(struct file_lock *); | |
890 | }; | |
891 | ||
892 | struct lock_manager_operations { | |
893 | int (*lm_compare_owner)(struct file_lock *, struct file_lock *); | |
894 | unsigned long (*lm_owner_key)(struct file_lock *); | |
895 | void (*lm_get_owner)(struct file_lock *, struct file_lock *); | |
896 | void (*lm_put_owner)(struct file_lock *); | |
897 | void (*lm_notify)(struct file_lock *); /* unblock callback */ | |
898 | int (*lm_grant)(struct file_lock *, int); | |
899 | bool (*lm_break)(struct file_lock *); | |
900 | int (*lm_change)(struct file_lock **, int, struct list_head *); | |
901 | void (*lm_setup)(struct file_lock *, void **); | |
902 | }; | |
903 | ||
904 | struct lock_manager { | |
905 | struct list_head list; | |
906 | }; | |
907 | ||
908 | struct net; | |
909 | void locks_start_grace(struct net *, struct lock_manager *); | |
910 | void locks_end_grace(struct lock_manager *); | |
911 | int locks_in_grace(struct net *); | |
912 | ||
913 | /* that will die - we need it for nfs_lock_info */ | |
914 | #include <linux/nfs_fs_i.h> | |
915 | ||
916 | /* | |
917 | * struct file_lock represents a generic "file lock". It's used to represent | |
918 | * POSIX byte range locks, BSD (flock) locks, and leases. It's important to | |
919 | * note that the same struct is used to represent both a request for a lock and | |
920 | * the lock itself, but the same object is never used for both. | |
921 | * | |
922 | * FIXME: should we create a separate "struct lock_request" to help distinguish | |
923 | * these two uses? | |
924 | * | |
925 | * The i_flock list is ordered by: | |
926 | * | |
927 | * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX | |
928 | * 2) lock owner | |
929 | * 3) lock range start | |
930 | * 4) lock range end | |
931 | * | |
932 | * Obviously, the last two criteria only matter for POSIX locks. | |
933 | */ | |
934 | struct file_lock { | |
935 | struct file_lock *fl_next; /* singly linked list for this inode */ | |
936 | struct hlist_node fl_link; /* node in global lists */ | |
937 | struct list_head fl_block; /* circular list of blocked processes */ | |
938 | fl_owner_t fl_owner; | |
939 | unsigned int fl_flags; | |
940 | unsigned char fl_type; | |
941 | unsigned int fl_pid; | |
942 | int fl_link_cpu; /* what cpu's list is this on? */ | |
943 | struct pid *fl_nspid; | |
944 | wait_queue_head_t fl_wait; | |
945 | struct file *fl_file; | |
946 | loff_t fl_start; | |
947 | loff_t fl_end; | |
948 | ||
949 | struct fasync_struct * fl_fasync; /* for lease break notifications */ | |
950 | /* for lease breaks: */ | |
951 | unsigned long fl_break_time; | |
952 | unsigned long fl_downgrade_time; | |
953 | ||
954 | const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ | |
955 | const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ | |
956 | union { | |
957 | struct nfs_lock_info nfs_fl; | |
958 | struct nfs4_lock_info nfs4_fl; | |
959 | struct { | |
960 | struct list_head link; /* link in AFS vnode's pending_locks list */ | |
961 | int state; /* state of grant or error if -ve */ | |
962 | } afs; | |
963 | } fl_u; | |
964 | }; | |
965 | ||
966 | /* The following constant reflects the upper bound of the file/locking space */ | |
967 | #ifndef OFFSET_MAX | |
968 | #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) | |
969 | #define OFFSET_MAX INT_LIMIT(loff_t) | |
970 | #define OFFT_OFFSET_MAX INT_LIMIT(off_t) | |
971 | #endif | |
972 | ||
973 | #include <linux/fcntl.h> | |
974 | ||
975 | extern void send_sigio(struct fown_struct *fown, int fd, int band); | |
976 | ||
977 | #ifdef CONFIG_FILE_LOCKING | |
978 | extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *); | |
979 | extern int fcntl_setlk(unsigned int, struct file *, unsigned int, | |
980 | struct flock __user *); | |
981 | ||
982 | #if BITS_PER_LONG == 32 | |
983 | extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *); | |
984 | extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, | |
985 | struct flock64 __user *); | |
986 | #endif | |
987 | ||
988 | extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); | |
989 | extern int fcntl_getlease(struct file *filp); | |
990 | ||
991 | /* fs/locks.c */ | |
992 | void locks_free_lock(struct file_lock *fl); | |
993 | extern void locks_init_lock(struct file_lock *); | |
994 | extern struct file_lock * locks_alloc_lock(void); | |
995 | extern void locks_copy_lock(struct file_lock *, struct file_lock *); | |
996 | extern void locks_copy_conflock(struct file_lock *, struct file_lock *); | |
997 | extern void locks_remove_posix(struct file *, fl_owner_t); | |
998 | extern void locks_remove_file(struct file *); | |
999 | extern void locks_release_private(struct file_lock *); | |
1000 | extern void posix_test_lock(struct file *, struct file_lock *); | |
1001 | extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); | |
1002 | extern int posix_lock_file_wait(struct file *, struct file_lock *); | |
1003 | extern int posix_unblock_lock(struct file_lock *); | |
1004 | extern int vfs_test_lock(struct file *, struct file_lock *); | |
1005 | extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); | |
1006 | extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); | |
1007 | extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl); | |
1008 | extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type); | |
1009 | extern void lease_get_mtime(struct inode *, struct timespec *time); | |
1010 | extern int generic_setlease(struct file *, long, struct file_lock **, void **priv); | |
1011 | extern int vfs_setlease(struct file *, long, struct file_lock **, void **); | |
1012 | extern int lease_modify(struct file_lock **, int, struct list_head *); | |
1013 | #else /* !CONFIG_FILE_LOCKING */ | |
1014 | static inline int fcntl_getlk(struct file *file, unsigned int cmd, | |
1015 | struct flock __user *user) | |
1016 | { | |
1017 | return -EINVAL; | |
1018 | } | |
1019 | ||
1020 | static inline int fcntl_setlk(unsigned int fd, struct file *file, | |
1021 | unsigned int cmd, struct flock __user *user) | |
1022 | { | |
1023 | return -EACCES; | |
1024 | } | |
1025 | ||
1026 | #if BITS_PER_LONG == 32 | |
1027 | static inline int fcntl_getlk64(struct file *file, unsigned int cmd, | |
1028 | struct flock64 __user *user) | |
1029 | { | |
1030 | return -EINVAL; | |
1031 | } | |
1032 | ||
1033 | static inline int fcntl_setlk64(unsigned int fd, struct file *file, | |
1034 | unsigned int cmd, struct flock64 __user *user) | |
1035 | { | |
1036 | return -EACCES; | |
1037 | } | |
1038 | #endif | |
1039 | static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) | |
1040 | { | |
1041 | return -EINVAL; | |
1042 | } | |
1043 | ||
1044 | static inline int fcntl_getlease(struct file *filp) | |
1045 | { | |
1046 | return F_UNLCK; | |
1047 | } | |
1048 | ||
1049 | static inline void locks_init_lock(struct file_lock *fl) | |
1050 | { | |
1051 | return; | |
1052 | } | |
1053 | ||
1054 | static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl) | |
1055 | { | |
1056 | return; | |
1057 | } | |
1058 | ||
1059 | static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) | |
1060 | { | |
1061 | return; | |
1062 | } | |
1063 | ||
1064 | static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) | |
1065 | { | |
1066 | return; | |
1067 | } | |
1068 | ||
1069 | static inline void locks_remove_file(struct file *filp) | |
1070 | { | |
1071 | return; | |
1072 | } | |
1073 | ||
1074 | static inline void posix_test_lock(struct file *filp, struct file_lock *fl) | |
1075 | { | |
1076 | return; | |
1077 | } | |
1078 | ||
1079 | static inline int posix_lock_file(struct file *filp, struct file_lock *fl, | |
1080 | struct file_lock *conflock) | |
1081 | { | |
1082 | return -ENOLCK; | |
1083 | } | |
1084 | ||
1085 | static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl) | |
1086 | { | |
1087 | return -ENOLCK; | |
1088 | } | |
1089 | ||
1090 | static inline int posix_unblock_lock(struct file_lock *waiter) | |
1091 | { | |
1092 | return -ENOENT; | |
1093 | } | |
1094 | ||
1095 | static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) | |
1096 | { | |
1097 | return 0; | |
1098 | } | |
1099 | ||
1100 | static inline int vfs_lock_file(struct file *filp, unsigned int cmd, | |
1101 | struct file_lock *fl, struct file_lock *conf) | |
1102 | { | |
1103 | return -ENOLCK; | |
1104 | } | |
1105 | ||
1106 | static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) | |
1107 | { | |
1108 | return 0; | |
1109 | } | |
1110 | ||
1111 | static inline int flock_lock_file_wait(struct file *filp, | |
1112 | struct file_lock *request) | |
1113 | { | |
1114 | return -ENOLCK; | |
1115 | } | |
1116 | ||
1117 | static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type) | |
1118 | { | |
1119 | return 0; | |
1120 | } | |
1121 | ||
1122 | static inline void lease_get_mtime(struct inode *inode, struct timespec *time) | |
1123 | { | |
1124 | return; | |
1125 | } | |
1126 | ||
1127 | static inline int generic_setlease(struct file *filp, long arg, | |
1128 | struct file_lock **flp, void **priv) | |
1129 | { | |
1130 | return -EINVAL; | |
1131 | } | |
1132 | ||
1133 | static inline int vfs_setlease(struct file *filp, long arg, | |
1134 | struct file_lock **lease, void **priv) | |
1135 | { | |
1136 | return -EINVAL; | |
1137 | } | |
1138 | ||
1139 | static inline int lease_modify(struct file_lock **before, int arg, | |
1140 | struct list_head *dispose) | |
1141 | { | |
1142 | return -EINVAL; | |
1143 | } | |
1144 | #endif /* !CONFIG_FILE_LOCKING */ | |
1145 | ||
1146 | ||
1147 | struct fasync_struct { | |
1148 | spinlock_t fa_lock; | |
1149 | int magic; | |
1150 | int fa_fd; | |
1151 | struct fasync_struct *fa_next; /* singly linked list */ | |
1152 | struct file *fa_file; | |
1153 | struct rcu_head fa_rcu; | |
1154 | }; | |
1155 | ||
1156 | #define FASYNC_MAGIC 0x4601 | |
1157 | ||
1158 | /* SMP safe fasync helpers: */ | |
1159 | extern int fasync_helper(int, struct file *, int, struct fasync_struct **); | |
1160 | extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); | |
1161 | extern int fasync_remove_entry(struct file *, struct fasync_struct **); | |
1162 | extern struct fasync_struct *fasync_alloc(void); | |
1163 | extern void fasync_free(struct fasync_struct *); | |
1164 | ||
1165 | /* can be called from interrupts */ | |
1166 | extern void kill_fasync(struct fasync_struct **, int, int); | |
1167 | ||
1168 | extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); | |
1169 | extern void f_setown(struct file *filp, unsigned long arg, int force); | |
1170 | extern void f_delown(struct file *filp); | |
1171 | extern pid_t f_getown(struct file *filp); | |
1172 | extern int send_sigurg(struct fown_struct *fown); | |
1173 | ||
1174 | struct mm_struct; | |
1175 | ||
1176 | /* | |
1177 | * Umount options | |
1178 | */ | |
1179 | ||
1180 | #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ | |
1181 | #define MNT_DETACH 0x00000002 /* Just detach from the tree */ | |
1182 | #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ | |
1183 | #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ | |
1184 | #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ | |
1185 | ||
1186 | extern struct list_head super_blocks; | |
1187 | extern spinlock_t sb_lock; | |
1188 | ||
1189 | /* Possible states of 'frozen' field */ | |
1190 | enum { | |
1191 | SB_UNFROZEN = 0, /* FS is unfrozen */ | |
1192 | SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ | |
1193 | SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ | |
1194 | SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop | |
1195 | * internal threads if needed) */ | |
1196 | SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ | |
1197 | }; | |
1198 | ||
1199 | #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) | |
1200 | ||
1201 | struct sb_writers { | |
1202 | /* Counters for counting writers at each level */ | |
1203 | struct percpu_counter counter[SB_FREEZE_LEVELS]; | |
1204 | wait_queue_head_t wait; /* queue for waiting for | |
1205 | writers / faults to finish */ | |
1206 | int frozen; /* Is sb frozen? */ | |
1207 | wait_queue_head_t wait_unfrozen; /* queue for waiting for | |
1208 | sb to be thawed */ | |
1209 | #ifdef CONFIG_DEBUG_LOCK_ALLOC | |
1210 | struct lockdep_map lock_map[SB_FREEZE_LEVELS]; | |
1211 | #endif | |
1212 | }; | |
1213 | ||
1214 | struct super_block { | |
1215 | struct list_head s_list; /* Keep this first */ | |
1216 | dev_t s_dev; /* search index; _not_ kdev_t */ | |
1217 | unsigned char s_blocksize_bits; | |
1218 | unsigned long s_blocksize; | |
1219 | loff_t s_maxbytes; /* Max file size */ | |
1220 | struct file_system_type *s_type; | |
1221 | const struct super_operations *s_op; | |
1222 | const struct dquot_operations *dq_op; | |
1223 | const struct quotactl_ops *s_qcop; | |
1224 | const struct export_operations *s_export_op; | |
1225 | unsigned long s_flags; | |
1226 | unsigned long s_magic; | |
1227 | struct dentry *s_root; | |
1228 | struct rw_semaphore s_umount; | |
1229 | int s_count; | |
1230 | atomic_t s_active; | |
1231 | #ifdef CONFIG_SECURITY | |
1232 | void *s_security; | |
1233 | #endif | |
1234 | const struct xattr_handler **s_xattr; | |
1235 | ||
1236 | struct list_head s_inodes; /* all inodes */ | |
1237 | struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */ | |
1238 | struct list_head s_mounts; /* list of mounts; _not_ for fs use */ | |
1239 | struct block_device *s_bdev; | |
1240 | struct backing_dev_info *s_bdi; | |
1241 | struct mtd_info *s_mtd; | |
1242 | struct hlist_node s_instances; | |
1243 | unsigned int s_quota_types; /* Bitmask of supported quota types */ | |
1244 | struct quota_info s_dquot; /* Diskquota specific options */ | |
1245 | ||
1246 | struct sb_writers s_writers; | |
1247 | ||
1248 | char s_id[32]; /* Informational name */ | |
1249 | u8 s_uuid[16]; /* UUID */ | |
1250 | ||
1251 | void *s_fs_info; /* Filesystem private info */ | |
1252 | unsigned int s_max_links; | |
1253 | fmode_t s_mode; | |
1254 | ||
1255 | /* Granularity of c/m/atime in ns. | |
1256 | Cannot be worse than a second */ | |
1257 | u32 s_time_gran; | |
1258 | ||
1259 | /* | |
1260 | * The next field is for VFS *only*. No filesystems have any business | |
1261 | * even looking at it. You had been warned. | |
1262 | */ | |
1263 | struct mutex s_vfs_rename_mutex; /* Kludge */ | |
1264 | ||
1265 | /* | |
1266 | * Filesystem subtype. If non-empty the filesystem type field | |
1267 | * in /proc/mounts will be "type.subtype" | |
1268 | */ | |
1269 | char *s_subtype; | |
1270 | ||
1271 | /* | |
1272 | * Saved mount options for lazy filesystems using | |
1273 | * generic_show_options() | |
1274 | */ | |
1275 | char __rcu *s_options; | |
1276 | const struct dentry_operations *s_d_op; /* default d_op for dentries */ | |
1277 | ||
1278 | /* | |
1279 | * Saved pool identifier for cleancache (-1 means none) | |
1280 | */ | |
1281 | int cleancache_poolid; | |
1282 | ||
1283 | struct shrinker s_shrink; /* per-sb shrinker handle */ | |
1284 | ||
1285 | /* Number of inodes with nlink == 0 but still referenced */ | |
1286 | atomic_long_t s_remove_count; | |
1287 | ||
1288 | /* Being remounted read-only */ | |
1289 | int s_readonly_remount; | |
1290 | ||
1291 | /* AIO completions deferred from interrupt context */ | |
1292 | struct workqueue_struct *s_dio_done_wq; | |
1293 | struct hlist_head s_pins; | |
1294 | ||
1295 | /* | |
1296 | * Keep the lru lists last in the structure so they always sit on their | |
1297 | * own individual cachelines. | |
1298 | */ | |
1299 | struct list_lru s_dentry_lru ____cacheline_aligned_in_smp; | |
1300 | struct list_lru s_inode_lru ____cacheline_aligned_in_smp; | |
1301 | struct rcu_head rcu; | |
1302 | ||
1303 | /* | |
1304 | * Indicates how deep in a filesystem stack this SB is | |
1305 | */ | |
1306 | int s_stack_depth; | |
1307 | }; | |
1308 | ||
1309 | extern struct timespec current_fs_time(struct super_block *sb); | |
1310 | ||
1311 | /* | |
1312 | * Snapshotting support. | |
1313 | */ | |
1314 | ||
1315 | void __sb_end_write(struct super_block *sb, int level); | |
1316 | int __sb_start_write(struct super_block *sb, int level, bool wait); | |
1317 | ||
1318 | /** | |
1319 | * sb_end_write - drop write access to a superblock | |
1320 | * @sb: the super we wrote to | |
1321 | * | |
1322 | * Decrement number of writers to the filesystem. Wake up possible waiters | |
1323 | * wanting to freeze the filesystem. | |
1324 | */ | |
1325 | static inline void sb_end_write(struct super_block *sb) | |
1326 | { | |
1327 | __sb_end_write(sb, SB_FREEZE_WRITE); | |
1328 | } | |
1329 | ||
1330 | /** | |
1331 | * sb_end_pagefault - drop write access to a superblock from a page fault | |
1332 | * @sb: the super we wrote to | |
1333 | * | |
1334 | * Decrement number of processes handling write page fault to the filesystem. | |
1335 | * Wake up possible waiters wanting to freeze the filesystem. | |
1336 | */ | |
1337 | static inline void sb_end_pagefault(struct super_block *sb) | |
1338 | { | |
1339 | __sb_end_write(sb, SB_FREEZE_PAGEFAULT); | |
1340 | } | |
1341 | ||
1342 | /** | |
1343 | * sb_end_intwrite - drop write access to a superblock for internal fs purposes | |
1344 | * @sb: the super we wrote to | |
1345 | * | |
1346 | * Decrement fs-internal number of writers to the filesystem. Wake up possible | |
1347 | * waiters wanting to freeze the filesystem. | |
1348 | */ | |
1349 | static inline void sb_end_intwrite(struct super_block *sb) | |
1350 | { | |
1351 | __sb_end_write(sb, SB_FREEZE_FS); | |
1352 | } | |
1353 | ||
1354 | /** | |
1355 | * sb_start_write - get write access to a superblock | |
1356 | * @sb: the super we write to | |
1357 | * | |
1358 | * When a process wants to write data or metadata to a file system (i.e. dirty | |
1359 | * a page or an inode), it should embed the operation in a sb_start_write() - | |
1360 | * sb_end_write() pair to get exclusion against file system freezing. This | |
1361 | * function increments number of writers preventing freezing. If the file | |
1362 | * system is already frozen, the function waits until the file system is | |
1363 | * thawed. | |
1364 | * | |
1365 | * Since freeze protection behaves as a lock, users have to preserve | |
1366 | * ordering of freeze protection and other filesystem locks. Generally, | |
1367 | * freeze protection should be the outermost lock. In particular, we have: | |
1368 | * | |
1369 | * sb_start_write | |
1370 | * -> i_mutex (write path, truncate, directory ops, ...) | |
1371 | * -> s_umount (freeze_super, thaw_super) | |
1372 | */ | |
1373 | static inline void sb_start_write(struct super_block *sb) | |
1374 | { | |
1375 | __sb_start_write(sb, SB_FREEZE_WRITE, true); | |
1376 | } | |
1377 | ||
1378 | static inline int sb_start_write_trylock(struct super_block *sb) | |
1379 | { | |
1380 | return __sb_start_write(sb, SB_FREEZE_WRITE, false); | |
1381 | } | |
1382 | ||
1383 | /** | |
1384 | * sb_start_pagefault - get write access to a superblock from a page fault | |
1385 | * @sb: the super we write to | |
1386 | * | |
1387 | * When a process starts handling write page fault, it should embed the | |
1388 | * operation into sb_start_pagefault() - sb_end_pagefault() pair to get | |
1389 | * exclusion against file system freezing. This is needed since the page fault | |
1390 | * is going to dirty a page. This function increments number of running page | |
1391 | * faults preventing freezing. If the file system is already frozen, the | |
1392 | * function waits until the file system is thawed. | |
1393 | * | |
1394 | * Since page fault freeze protection behaves as a lock, users have to preserve | |
1395 | * ordering of freeze protection and other filesystem locks. It is advised to | |
1396 | * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault | |
1397 | * handling code implies lock dependency: | |
1398 | * | |
1399 | * mmap_sem | |
1400 | * -> sb_start_pagefault | |
1401 | */ | |
1402 | static inline void sb_start_pagefault(struct super_block *sb) | |
1403 | { | |
1404 | __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true); | |
1405 | } | |
1406 | ||
1407 | /* | |
1408 | * sb_start_intwrite - get write access to a superblock for internal fs purposes | |
1409 | * @sb: the super we write to | |
1410 | * | |
1411 | * This is the third level of protection against filesystem freezing. It is | |
1412 | * free for use by a filesystem. The only requirement is that it must rank | |
1413 | * below sb_start_pagefault. | |
1414 | * | |
1415 | * For example filesystem can call sb_start_intwrite() when starting a | |
1416 | * transaction which somewhat eases handling of freezing for internal sources | |
1417 | * of filesystem changes (internal fs threads, discarding preallocation on file | |
1418 | * close, etc.). | |
1419 | */ | |
1420 | static inline void sb_start_intwrite(struct super_block *sb) | |
1421 | { | |
1422 | __sb_start_write(sb, SB_FREEZE_FS, true); | |
1423 | } | |
1424 | ||
1425 | ||
1426 | extern bool inode_owner_or_capable(const struct inode *inode); | |
1427 | ||
1428 | /* | |
1429 | * VFS helper functions.. | |
1430 | */ | |
1431 | extern int vfs_create(struct inode *, struct dentry *, umode_t, bool); | |
1432 | extern int vfs_mkdir(struct inode *, struct dentry *, umode_t); | |
1433 | extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t); | |
1434 | extern int vfs_symlink(struct inode *, struct dentry *, const char *); | |
1435 | extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **); | |
1436 | extern int vfs_rmdir(struct inode *, struct dentry *); | |
1437 | extern int vfs_unlink(struct inode *, struct dentry *, struct inode **); | |
1438 | extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int); | |
1439 | extern int vfs_whiteout(struct inode *, struct dentry *); | |
1440 | ||
1441 | /* | |
1442 | * VFS dentry helper functions. | |
1443 | */ | |
1444 | extern void dentry_unhash(struct dentry *dentry); | |
1445 | ||
1446 | /* | |
1447 | * VFS file helper functions. | |
1448 | */ | |
1449 | extern void inode_init_owner(struct inode *inode, const struct inode *dir, | |
1450 | umode_t mode); | |
1451 | /* | |
1452 | * VFS FS_IOC_FIEMAP helper definitions. | |
1453 | */ | |
1454 | struct fiemap_extent_info { | |
1455 | unsigned int fi_flags; /* Flags as passed from user */ | |
1456 | unsigned int fi_extents_mapped; /* Number of mapped extents */ | |
1457 | unsigned int fi_extents_max; /* Size of fiemap_extent array */ | |
1458 | struct fiemap_extent __user *fi_extents_start; /* Start of | |
1459 | fiemap_extent array */ | |
1460 | }; | |
1461 | int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical, | |
1462 | u64 phys, u64 len, u32 flags); | |
1463 | int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags); | |
1464 | ||
1465 | /* | |
1466 | * File types | |
1467 | * | |
1468 | * NOTE! These match bits 12..15 of stat.st_mode | |
1469 | * (ie "(i_mode >> 12) & 15"). | |
1470 | */ | |
1471 | #define DT_UNKNOWN 0 | |
1472 | #define DT_FIFO 1 | |
1473 | #define DT_CHR 2 | |
1474 | #define DT_DIR 4 | |
1475 | #define DT_BLK 6 | |
1476 | #define DT_REG 8 | |
1477 | #define DT_LNK 10 | |
1478 | #define DT_SOCK 12 | |
1479 | #define DT_WHT 14 | |
1480 | ||
1481 | /* | |
1482 | * This is the "filldir" function type, used by readdir() to let | |
1483 | * the kernel specify what kind of dirent layout it wants to have. | |
1484 | * This allows the kernel to read directories into kernel space or | |
1485 | * to have different dirent layouts depending on the binary type. | |
1486 | */ | |
1487 | struct dir_context; | |
1488 | typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, | |
1489 | unsigned); | |
1490 | ||
1491 | struct dir_context { | |
1492 | const filldir_t actor; | |
1493 | loff_t pos; | |
1494 | }; | |
1495 | ||
1496 | struct block_device_operations; | |
1497 | ||
1498 | /* These macros are for out of kernel modules to test that | |
1499 | * the kernel supports the unlocked_ioctl and compat_ioctl | |
1500 | * fields in struct file_operations. */ | |
1501 | #define HAVE_COMPAT_IOCTL 1 | |
1502 | #define HAVE_UNLOCKED_IOCTL 1 | |
1503 | ||
1504 | /* | |
1505 | * These flags let !MMU mmap() govern direct device mapping vs immediate | |
1506 | * copying more easily for MAP_PRIVATE, especially for ROM filesystems. | |
1507 | * | |
1508 | * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) | |
1509 | * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) | |
1510 | * NOMMU_MAP_READ: Can be mapped for reading | |
1511 | * NOMMU_MAP_WRITE: Can be mapped for writing | |
1512 | * NOMMU_MAP_EXEC: Can be mapped for execution | |
1513 | */ | |
1514 | #define NOMMU_MAP_COPY 0x00000001 | |
1515 | #define NOMMU_MAP_DIRECT 0x00000008 | |
1516 | #define NOMMU_MAP_READ VM_MAYREAD | |
1517 | #define NOMMU_MAP_WRITE VM_MAYWRITE | |
1518 | #define NOMMU_MAP_EXEC VM_MAYEXEC | |
1519 | ||
1520 | #define NOMMU_VMFLAGS \ | |
1521 | (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) | |
1522 | ||
1523 | ||
1524 | struct iov_iter; | |
1525 | ||
1526 | struct file_operations { | |
1527 | struct module *owner; | |
1528 | loff_t (*llseek) (struct file *, loff_t, int); | |
1529 | ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); | |
1530 | ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); | |
1531 | ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t); | |
1532 | ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t); | |
1533 | ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); | |
1534 | ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); | |
1535 | int (*iterate) (struct file *, struct dir_context *); | |
1536 | unsigned int (*poll) (struct file *, struct poll_table_struct *); | |
1537 | long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); | |
1538 | long (*compat_ioctl) (struct file *, unsigned int, unsigned long); | |
1539 | int (*mmap) (struct file *, struct vm_area_struct *); | |
1540 | void (*mremap)(struct file *, struct vm_area_struct *); | |
1541 | int (*open) (struct inode *, struct file *); | |
1542 | int (*flush) (struct file *, fl_owner_t id); | |
1543 | int (*release) (struct inode *, struct file *); | |
1544 | int (*fsync) (struct file *, loff_t, loff_t, int datasync); | |
1545 | int (*aio_fsync) (struct kiocb *, int datasync); | |
1546 | int (*fasync) (int, struct file *, int); | |
1547 | int (*lock) (struct file *, int, struct file_lock *); | |
1548 | ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); | |
1549 | unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1550 | int (*check_flags)(int); | |
1551 | int (*flock) (struct file *, int, struct file_lock *); | |
1552 | ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); | |
1553 | ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); | |
1554 | int (*setlease)(struct file *, long, struct file_lock **, void **); | |
1555 | long (*fallocate)(struct file *file, int mode, loff_t offset, | |
1556 | loff_t len); | |
1557 | void (*show_fdinfo)(struct seq_file *m, struct file *f); | |
1558 | #ifndef CONFIG_MMU | |
1559 | unsigned (*mmap_capabilities)(struct file *); | |
1560 | #endif | |
1561 | }; | |
1562 | ||
1563 | struct inode_operations { | |
1564 | struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); | |
1565 | void * (*follow_link) (struct dentry *, struct nameidata *); | |
1566 | int (*permission) (struct inode *, int); | |
1567 | struct posix_acl * (*get_acl)(struct inode *, int); | |
1568 | ||
1569 | int (*readlink) (struct dentry *, char __user *,int); | |
1570 | void (*put_link) (struct dentry *, struct nameidata *, void *); | |
1571 | ||
1572 | int (*create) (struct inode *,struct dentry *, umode_t, bool); | |
1573 | int (*link) (struct dentry *,struct inode *,struct dentry *); | |
1574 | int (*unlink) (struct inode *,struct dentry *); | |
1575 | int (*symlink) (struct inode *,struct dentry *,const char *); | |
1576 | int (*mkdir) (struct inode *,struct dentry *,umode_t); | |
1577 | int (*rmdir) (struct inode *,struct dentry *); | |
1578 | int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t); | |
1579 | int (*rename) (struct inode *, struct dentry *, | |
1580 | struct inode *, struct dentry *); | |
1581 | int (*rename2) (struct inode *, struct dentry *, | |
1582 | struct inode *, struct dentry *, unsigned int); | |
1583 | int (*setattr) (struct dentry *, struct iattr *); | |
1584 | int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *); | |
1585 | int (*setxattr) (struct dentry *, const char *,const void *,size_t,int); | |
1586 | ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t); | |
1587 | ssize_t (*listxattr) (struct dentry *, char *, size_t); | |
1588 | int (*removexattr) (struct dentry *, const char *); | |
1589 | int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, | |
1590 | u64 len); | |
1591 | int (*update_time)(struct inode *, struct timespec *, int); | |
1592 | int (*atomic_open)(struct inode *, struct dentry *, | |
1593 | struct file *, unsigned open_flag, | |
1594 | umode_t create_mode, int *opened); | |
1595 | int (*tmpfile) (struct inode *, struct dentry *, umode_t); | |
1596 | int (*set_acl)(struct inode *, struct posix_acl *, int); | |
1597 | ||
1598 | /* WARNING: probably going away soon, do not use! */ | |
1599 | int (*dentry_open)(struct dentry *, struct file *, const struct cred *); | |
1600 | } ____cacheline_aligned; | |
1601 | ||
1602 | ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, | |
1603 | unsigned long nr_segs, unsigned long fast_segs, | |
1604 | struct iovec *fast_pointer, | |
1605 | struct iovec **ret_pointer); | |
1606 | ||
1607 | extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *); | |
1608 | extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); | |
1609 | extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); | |
1610 | extern ssize_t vfs_readv(struct file *, const struct iovec __user *, | |
1611 | unsigned long, loff_t *); | |
1612 | extern ssize_t vfs_writev(struct file *, const struct iovec __user *, | |
1613 | unsigned long, loff_t *); | |
1614 | ||
1615 | struct super_operations { | |
1616 | struct inode *(*alloc_inode)(struct super_block *sb); | |
1617 | void (*destroy_inode)(struct inode *); | |
1618 | ||
1619 | void (*dirty_inode) (struct inode *, int flags); | |
1620 | int (*write_inode) (struct inode *, struct writeback_control *wbc); | |
1621 | int (*drop_inode) (struct inode *); | |
1622 | void (*evict_inode) (struct inode *); | |
1623 | void (*put_super) (struct super_block *); | |
1624 | int (*sync_fs)(struct super_block *sb, int wait); | |
1625 | int (*freeze_super) (struct super_block *); | |
1626 | int (*freeze_fs) (struct super_block *); | |
1627 | int (*thaw_super) (struct super_block *); | |
1628 | int (*unfreeze_fs) (struct super_block *); | |
1629 | int (*statfs) (struct dentry *, struct kstatfs *); | |
1630 | int (*remount_fs) (struct super_block *, int *, char *); | |
1631 | void (*umount_begin) (struct super_block *); | |
1632 | ||
1633 | int (*show_options)(struct seq_file *, struct dentry *); | |
1634 | int (*show_devname)(struct seq_file *, struct dentry *); | |
1635 | int (*show_path)(struct seq_file *, struct dentry *); | |
1636 | int (*show_stats)(struct seq_file *, struct dentry *); | |
1637 | #ifdef CONFIG_QUOTA | |
1638 | ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); | |
1639 | ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); | |
1640 | struct dquot **(*get_dquots)(struct inode *); | |
1641 | #endif | |
1642 | int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t); | |
1643 | long (*nr_cached_objects)(struct super_block *, int); | |
1644 | long (*free_cached_objects)(struct super_block *, long, int); | |
1645 | }; | |
1646 | ||
1647 | /* | |
1648 | * Inode flags - they have no relation to superblock flags now | |
1649 | */ | |
1650 | #define S_SYNC 1 /* Writes are synced at once */ | |
1651 | #define S_NOATIME 2 /* Do not update access times */ | |
1652 | #define S_APPEND 4 /* Append-only file */ | |
1653 | #define S_IMMUTABLE 8 /* Immutable file */ | |
1654 | #define S_DEAD 16 /* removed, but still open directory */ | |
1655 | #define S_NOQUOTA 32 /* Inode is not counted to quota */ | |
1656 | #define S_DIRSYNC 64 /* Directory modifications are synchronous */ | |
1657 | #define S_NOCMTIME 128 /* Do not update file c/mtime */ | |
1658 | #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ | |
1659 | #define S_PRIVATE 512 /* Inode is fs-internal */ | |
1660 | #define S_IMA 1024 /* Inode has an associated IMA struct */ | |
1661 | #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */ | |
1662 | #define S_NOSEC 4096 /* no suid or xattr security attributes */ | |
1663 | ||
1664 | /* | |
1665 | * Note that nosuid etc flags are inode-specific: setting some file-system | |
1666 | * flags just means all the inodes inherit those flags by default. It might be | |
1667 | * possible to override it selectively if you really wanted to with some | |
1668 | * ioctl() that is not currently implemented. | |
1669 | * | |
1670 | * Exception: MS_RDONLY is always applied to the entire file system. | |
1671 | * | |
1672 | * Unfortunately, it is possible to change a filesystems flags with it mounted | |
1673 | * with files in use. This means that all of the inodes will not have their | |
1674 | * i_flags updated. Hence, i_flags no longer inherit the superblock mount | |
1675 | * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org | |
1676 | */ | |
1677 | #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) | |
1678 | ||
1679 | #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY) | |
1680 | #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \ | |
1681 | ((inode)->i_flags & S_SYNC)) | |
1682 | #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \ | |
1683 | ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) | |
1684 | #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK) | |
1685 | #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME) | |
1686 | #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION) | |
1687 | ||
1688 | #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) | |
1689 | #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) | |
1690 | #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) | |
1691 | #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL) | |
1692 | ||
1693 | #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) | |
1694 | #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) | |
1695 | #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) | |
1696 | #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) | |
1697 | #define IS_IMA(inode) ((inode)->i_flags & S_IMA) | |
1698 | #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) | |
1699 | #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) | |
1700 | ||
1701 | #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ | |
1702 | (inode)->i_rdev == WHITEOUT_DEV) | |
1703 | ||
1704 | /* | |
1705 | * Inode state bits. Protected by inode->i_lock | |
1706 | * | |
1707 | * Three bits determine the dirty state of the inode, I_DIRTY_SYNC, | |
1708 | * I_DIRTY_DATASYNC and I_DIRTY_PAGES. | |
1709 | * | |
1710 | * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, | |
1711 | * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at | |
1712 | * various stages of removing an inode. | |
1713 | * | |
1714 | * Two bits are used for locking and completion notification, I_NEW and I_SYNC. | |
1715 | * | |
1716 | * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on | |
1717 | * fdatasync(). i_atime is the usual cause. | |
1718 | * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of | |
1719 | * these changes separately from I_DIRTY_SYNC so that we | |
1720 | * don't have to write inode on fdatasync() when only | |
1721 | * mtime has changed in it. | |
1722 | * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. | |
1723 | * I_NEW Serves as both a mutex and completion notification. | |
1724 | * New inodes set I_NEW. If two processes both create | |
1725 | * the same inode, one of them will release its inode and | |
1726 | * wait for I_NEW to be released before returning. | |
1727 | * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can | |
1728 | * also cause waiting on I_NEW, without I_NEW actually | |
1729 | * being set. find_inode() uses this to prevent returning | |
1730 | * nearly-dead inodes. | |
1731 | * I_WILL_FREE Must be set when calling write_inode_now() if i_count | |
1732 | * is zero. I_FREEING must be set when I_WILL_FREE is | |
1733 | * cleared. | |
1734 | * I_FREEING Set when inode is about to be freed but still has dirty | |
1735 | * pages or buffers attached or the inode itself is still | |
1736 | * dirty. | |
1737 | * I_CLEAR Added by clear_inode(). In this state the inode is | |
1738 | * clean and can be destroyed. Inode keeps I_FREEING. | |
1739 | * | |
1740 | * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are | |
1741 | * prohibited for many purposes. iget() must wait for | |
1742 | * the inode to be completely released, then create it | |
1743 | * anew. Other functions will just ignore such inodes, | |
1744 | * if appropriate. I_NEW is used for waiting. | |
1745 | * | |
1746 | * I_SYNC Writeback of inode is running. The bit is set during | |
1747 | * data writeback, and cleared with a wakeup on the bit | |
1748 | * address once it is done. The bit is also used to pin | |
1749 | * the inode in memory for flusher thread. | |
1750 | * | |
1751 | * I_REFERENCED Marks the inode as recently references on the LRU list. | |
1752 | * | |
1753 | * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). | |
1754 | * | |
1755 | * Q: What is the difference between I_WILL_FREE and I_FREEING? | |
1756 | */ | |
1757 | #define I_DIRTY_SYNC (1 << 0) | |
1758 | #define I_DIRTY_DATASYNC (1 << 1) | |
1759 | #define I_DIRTY_PAGES (1 << 2) | |
1760 | #define __I_NEW 3 | |
1761 | #define I_NEW (1 << __I_NEW) | |
1762 | #define I_WILL_FREE (1 << 4) | |
1763 | #define I_FREEING (1 << 5) | |
1764 | #define I_CLEAR (1 << 6) | |
1765 | #define __I_SYNC 7 | |
1766 | #define I_SYNC (1 << __I_SYNC) | |
1767 | #define I_REFERENCED (1 << 8) | |
1768 | #define __I_DIO_WAKEUP 9 | |
1769 | #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP) | |
1770 | #define I_LINKABLE (1 << 10) | |
1771 | ||
1772 | #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) | |
1773 | ||
1774 | extern void __mark_inode_dirty(struct inode *, int); | |
1775 | static inline void mark_inode_dirty(struct inode *inode) | |
1776 | { | |
1777 | __mark_inode_dirty(inode, I_DIRTY); | |
1778 | } | |
1779 | ||
1780 | static inline void mark_inode_dirty_sync(struct inode *inode) | |
1781 | { | |
1782 | __mark_inode_dirty(inode, I_DIRTY_SYNC); | |
1783 | } | |
1784 | ||
1785 | extern void inc_nlink(struct inode *inode); | |
1786 | extern void drop_nlink(struct inode *inode); | |
1787 | extern void clear_nlink(struct inode *inode); | |
1788 | extern void set_nlink(struct inode *inode, unsigned int nlink); | |
1789 | ||
1790 | static inline void inode_inc_link_count(struct inode *inode) | |
1791 | { | |
1792 | inc_nlink(inode); | |
1793 | mark_inode_dirty(inode); | |
1794 | } | |
1795 | ||
1796 | static inline void inode_dec_link_count(struct inode *inode) | |
1797 | { | |
1798 | drop_nlink(inode); | |
1799 | mark_inode_dirty(inode); | |
1800 | } | |
1801 | ||
1802 | /** | |
1803 | * inode_inc_iversion - increments i_version | |
1804 | * @inode: inode that need to be updated | |
1805 | * | |
1806 | * Every time the inode is modified, the i_version field will be incremented. | |
1807 | * The filesystem has to be mounted with i_version flag | |
1808 | */ | |
1809 | ||
1810 | static inline void inode_inc_iversion(struct inode *inode) | |
1811 | { | |
1812 | spin_lock(&inode->i_lock); | |
1813 | inode->i_version++; | |
1814 | spin_unlock(&inode->i_lock); | |
1815 | } | |
1816 | ||
1817 | enum file_time_flags { | |
1818 | S_ATIME = 1, | |
1819 | S_MTIME = 2, | |
1820 | S_CTIME = 4, | |
1821 | S_VERSION = 8, | |
1822 | }; | |
1823 | ||
1824 | extern void touch_atime(const struct path *); | |
1825 | static inline void file_accessed(struct file *file) | |
1826 | { | |
1827 | if (!(file->f_flags & O_NOATIME)) | |
1828 | touch_atime(&file->f_path); | |
1829 | } | |
1830 | ||
1831 | int sync_inode(struct inode *inode, struct writeback_control *wbc); | |
1832 | int sync_inode_metadata(struct inode *inode, int wait); | |
1833 | ||
1834 | struct file_system_type { | |
1835 | const char *name; | |
1836 | int fs_flags; | |
1837 | #define FS_REQUIRES_DEV 1 | |
1838 | #define FS_BINARY_MOUNTDATA 2 | |
1839 | #define FS_HAS_SUBTYPE 4 | |
1840 | #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ | |
1841 | #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */ | |
1842 | #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ | |
1843 | struct dentry *(*mount) (struct file_system_type *, int, | |
1844 | const char *, void *); | |
1845 | void (*kill_sb) (struct super_block *); | |
1846 | struct module *owner; | |
1847 | struct file_system_type * next; | |
1848 | struct hlist_head fs_supers; | |
1849 | ||
1850 | struct lock_class_key s_lock_key; | |
1851 | struct lock_class_key s_umount_key; | |
1852 | struct lock_class_key s_vfs_rename_key; | |
1853 | struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; | |
1854 | ||
1855 | struct lock_class_key i_lock_key; | |
1856 | struct lock_class_key i_mutex_key; | |
1857 | struct lock_class_key i_mutex_dir_key; | |
1858 | }; | |
1859 | ||
1860 | #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) | |
1861 | ||
1862 | extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags, | |
1863 | void *data, int (*fill_super)(struct super_block *, void *, int)); | |
1864 | extern struct dentry *mount_bdev(struct file_system_type *fs_type, | |
1865 | int flags, const char *dev_name, void *data, | |
1866 | int (*fill_super)(struct super_block *, void *, int)); | |
1867 | extern struct dentry *mount_single(struct file_system_type *fs_type, | |
1868 | int flags, void *data, | |
1869 | int (*fill_super)(struct super_block *, void *, int)); | |
1870 | extern struct dentry *mount_nodev(struct file_system_type *fs_type, | |
1871 | int flags, void *data, | |
1872 | int (*fill_super)(struct super_block *, void *, int)); | |
1873 | extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); | |
1874 | void generic_shutdown_super(struct super_block *sb); | |
1875 | void kill_block_super(struct super_block *sb); | |
1876 | void kill_anon_super(struct super_block *sb); | |
1877 | void kill_litter_super(struct super_block *sb); | |
1878 | void deactivate_super(struct super_block *sb); | |
1879 | void deactivate_locked_super(struct super_block *sb); | |
1880 | int set_anon_super(struct super_block *s, void *data); | |
1881 | int get_anon_bdev(dev_t *); | |
1882 | void free_anon_bdev(dev_t); | |
1883 | struct super_block *sget(struct file_system_type *type, | |
1884 | int (*test)(struct super_block *,void *), | |
1885 | int (*set)(struct super_block *,void *), | |
1886 | int flags, void *data); | |
1887 | extern struct dentry *mount_pseudo(struct file_system_type *, char *, | |
1888 | const struct super_operations *ops, | |
1889 | const struct dentry_operations *dops, | |
1890 | unsigned long); | |
1891 | ||
1892 | /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ | |
1893 | #define fops_get(fops) \ | |
1894 | (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) | |
1895 | #define fops_put(fops) \ | |
1896 | do { if (fops) module_put((fops)->owner); } while(0) | |
1897 | /* | |
1898 | * This one is to be used *ONLY* from ->open() instances. | |
1899 | * fops must be non-NULL, pinned down *and* module dependencies | |
1900 | * should be sufficient to pin the caller down as well. | |
1901 | */ | |
1902 | #define replace_fops(f, fops) \ | |
1903 | do { \ | |
1904 | struct file *__file = (f); \ | |
1905 | fops_put(__file->f_op); \ | |
1906 | BUG_ON(!(__file->f_op = (fops))); \ | |
1907 | } while(0) | |
1908 | ||
1909 | extern int register_filesystem(struct file_system_type *); | |
1910 | extern int unregister_filesystem(struct file_system_type *); | |
1911 | extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data); | |
1912 | #define kern_mount(type) kern_mount_data(type, NULL) | |
1913 | extern void kern_unmount(struct vfsmount *mnt); | |
1914 | extern int may_umount_tree(struct vfsmount *); | |
1915 | extern int may_umount(struct vfsmount *); | |
1916 | extern long do_mount(const char *, const char __user *, | |
1917 | const char *, unsigned long, void *); | |
1918 | extern struct vfsmount *collect_mounts(struct path *); | |
1919 | extern void drop_collected_mounts(struct vfsmount *); | |
1920 | extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *, | |
1921 | struct vfsmount *); | |
1922 | extern int vfs_statfs(struct path *, struct kstatfs *); | |
1923 | extern int user_statfs(const char __user *, struct kstatfs *); | |
1924 | extern int fd_statfs(int, struct kstatfs *); | |
1925 | extern int vfs_ustat(dev_t, struct kstatfs *); | |
1926 | extern int freeze_super(struct super_block *super); | |
1927 | extern int thaw_super(struct super_block *super); | |
1928 | extern bool our_mnt(struct vfsmount *mnt); | |
1929 | extern bool fs_fully_visible(struct file_system_type *); | |
1930 | ||
1931 | extern int current_umask(void); | |
1932 | ||
1933 | extern void ihold(struct inode * inode); | |
1934 | extern void iput(struct inode *); | |
1935 | ||
1936 | static inline struct inode *file_inode(const struct file *f) | |
1937 | { | |
1938 | return f->f_inode; | |
1939 | } | |
1940 | ||
1941 | /* /sys/fs */ | |
1942 | extern struct kobject *fs_kobj; | |
1943 | ||
1944 | #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK) | |
1945 | ||
1946 | #define FLOCK_VERIFY_READ 1 | |
1947 | #define FLOCK_VERIFY_WRITE 2 | |
1948 | ||
1949 | #ifdef CONFIG_FILE_LOCKING | |
1950 | extern int locks_mandatory_locked(struct file *); | |
1951 | extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t); | |
1952 | ||
1953 | /* | |
1954 | * Candidates for mandatory locking have the setgid bit set | |
1955 | * but no group execute bit - an otherwise meaningless combination. | |
1956 | */ | |
1957 | ||
1958 | static inline int __mandatory_lock(struct inode *ino) | |
1959 | { | |
1960 | return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; | |
1961 | } | |
1962 | ||
1963 | /* | |
1964 | * ... and these candidates should be on MS_MANDLOCK mounted fs, | |
1965 | * otherwise these will be advisory locks | |
1966 | */ | |
1967 | ||
1968 | static inline int mandatory_lock(struct inode *ino) | |
1969 | { | |
1970 | return IS_MANDLOCK(ino) && __mandatory_lock(ino); | |
1971 | } | |
1972 | ||
1973 | static inline int locks_verify_locked(struct file *file) | |
1974 | { | |
1975 | if (mandatory_lock(file_inode(file))) | |
1976 | return locks_mandatory_locked(file); | |
1977 | return 0; | |
1978 | } | |
1979 | ||
1980 | static inline int locks_verify_truncate(struct inode *inode, | |
1981 | struct file *filp, | |
1982 | loff_t size) | |
1983 | { | |
1984 | if (inode->i_flock && mandatory_lock(inode)) | |
1985 | return locks_mandatory_area( | |
1986 | FLOCK_VERIFY_WRITE, inode, filp, | |
1987 | size < inode->i_size ? size : inode->i_size, | |
1988 | (size < inode->i_size ? inode->i_size - size | |
1989 | : size - inode->i_size) | |
1990 | ); | |
1991 | return 0; | |
1992 | } | |
1993 | ||
1994 | static inline int break_lease(struct inode *inode, unsigned int mode) | |
1995 | { | |
1996 | /* | |
1997 | * Since this check is lockless, we must ensure that any refcounts | |
1998 | * taken are done before checking inode->i_flock. Otherwise, we could | |
1999 | * end up racing with tasks trying to set a new lease on this file. | |
2000 | */ | |
2001 | smp_mb(); | |
2002 | if (inode->i_flock) | |
2003 | return __break_lease(inode, mode, FL_LEASE); | |
2004 | return 0; | |
2005 | } | |
2006 | ||
2007 | static inline int break_deleg(struct inode *inode, unsigned int mode) | |
2008 | { | |
2009 | /* | |
2010 | * Since this check is lockless, we must ensure that any refcounts | |
2011 | * taken are done before checking inode->i_flock. Otherwise, we could | |
2012 | * end up racing with tasks trying to set a new lease on this file. | |
2013 | */ | |
2014 | smp_mb(); | |
2015 | if (inode->i_flock) | |
2016 | return __break_lease(inode, mode, FL_DELEG); | |
2017 | return 0; | |
2018 | } | |
2019 | ||
2020 | static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) | |
2021 | { | |
2022 | int ret; | |
2023 | ||
2024 | ret = break_deleg(inode, O_WRONLY|O_NONBLOCK); | |
2025 | if (ret == -EWOULDBLOCK && delegated_inode) { | |
2026 | *delegated_inode = inode; | |
2027 | ihold(inode); | |
2028 | } | |
2029 | return ret; | |
2030 | } | |
2031 | ||
2032 | static inline int break_deleg_wait(struct inode **delegated_inode) | |
2033 | { | |
2034 | int ret; | |
2035 | ||
2036 | ret = break_deleg(*delegated_inode, O_WRONLY); | |
2037 | iput(*delegated_inode); | |
2038 | *delegated_inode = NULL; | |
2039 | return ret; | |
2040 | } | |
2041 | ||
2042 | #else /* !CONFIG_FILE_LOCKING */ | |
2043 | static inline int locks_mandatory_locked(struct file *file) | |
2044 | { | |
2045 | return 0; | |
2046 | } | |
2047 | ||
2048 | static inline int locks_mandatory_area(int rw, struct inode *inode, | |
2049 | struct file *filp, loff_t offset, | |
2050 | size_t count) | |
2051 | { | |
2052 | return 0; | |
2053 | } | |
2054 | ||
2055 | static inline int __mandatory_lock(struct inode *inode) | |
2056 | { | |
2057 | return 0; | |
2058 | } | |
2059 | ||
2060 | static inline int mandatory_lock(struct inode *inode) | |
2061 | { | |
2062 | return 0; | |
2063 | } | |
2064 | ||
2065 | static inline int locks_verify_locked(struct file *file) | |
2066 | { | |
2067 | return 0; | |
2068 | } | |
2069 | ||
2070 | static inline int locks_verify_truncate(struct inode *inode, struct file *filp, | |
2071 | size_t size) | |
2072 | { | |
2073 | return 0; | |
2074 | } | |
2075 | ||
2076 | static inline int break_lease(struct inode *inode, unsigned int mode) | |
2077 | { | |
2078 | return 0; | |
2079 | } | |
2080 | ||
2081 | static inline int break_deleg(struct inode *inode, unsigned int mode) | |
2082 | { | |
2083 | return 0; | |
2084 | } | |
2085 | ||
2086 | static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) | |
2087 | { | |
2088 | return 0; | |
2089 | } | |
2090 | ||
2091 | static inline int break_deleg_wait(struct inode **delegated_inode) | |
2092 | { | |
2093 | BUG(); | |
2094 | return 0; | |
2095 | } | |
2096 | ||
2097 | #endif /* CONFIG_FILE_LOCKING */ | |
2098 | ||
2099 | /* fs/open.c */ | |
2100 | struct audit_names; | |
2101 | struct filename { | |
2102 | const char *name; /* pointer to actual string */ | |
2103 | const __user char *uptr; /* original userland pointer */ | |
2104 | struct audit_names *aname; | |
2105 | bool separate; /* should "name" be freed? */ | |
2106 | }; | |
2107 | ||
2108 | extern long vfs_truncate(struct path *, loff_t); | |
2109 | extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, | |
2110 | struct file *filp); | |
2111 | extern int vfs_fallocate(struct file *file, int mode, loff_t offset, | |
2112 | loff_t len); | |
2113 | extern long do_sys_open(int dfd, const char __user *filename, int flags, | |
2114 | umode_t mode); | |
2115 | extern struct file *file_open_name(struct filename *, int, umode_t); | |
2116 | extern struct file *filp_open(const char *, int, umode_t); | |
2117 | extern struct file *file_open_root(struct dentry *, struct vfsmount *, | |
2118 | const char *, int); | |
2119 | extern int vfs_open(const struct path *, struct file *, const struct cred *); | |
2120 | extern struct file * dentry_open(const struct path *, int, const struct cred *); | |
2121 | extern int filp_close(struct file *, fl_owner_t id); | |
2122 | ||
2123 | extern struct filename *getname_flags(const char __user *, int, int *); | |
2124 | extern struct filename *getname(const char __user *); | |
2125 | extern struct filename *getname_kernel(const char *); | |
2126 | ||
2127 | enum { | |
2128 | FILE_CREATED = 1, | |
2129 | FILE_OPENED = 2 | |
2130 | }; | |
2131 | extern int finish_open(struct file *file, struct dentry *dentry, | |
2132 | int (*open)(struct inode *, struct file *), | |
2133 | int *opened); | |
2134 | extern int finish_no_open(struct file *file, struct dentry *dentry); | |
2135 | ||
2136 | /* fs/ioctl.c */ | |
2137 | ||
2138 | extern int ioctl_preallocate(struct file *filp, void __user *argp); | |
2139 | ||
2140 | /* fs/dcache.c */ | |
2141 | extern void __init vfs_caches_init_early(void); | |
2142 | extern void __init vfs_caches_init(unsigned long); | |
2143 | ||
2144 | extern struct kmem_cache *names_cachep; | |
2145 | ||
2146 | extern void final_putname(struct filename *name); | |
2147 | ||
2148 | #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) | |
2149 | #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) | |
2150 | #ifndef CONFIG_AUDITSYSCALL | |
2151 | #define putname(name) final_putname(name) | |
2152 | #else | |
2153 | extern void putname(struct filename *name); | |
2154 | #endif | |
2155 | ||
2156 | #ifdef CONFIG_BLOCK | |
2157 | extern int register_blkdev(unsigned int, const char *); | |
2158 | extern void unregister_blkdev(unsigned int, const char *); | |
2159 | extern struct block_device *bdget(dev_t); | |
2160 | extern struct block_device *bdgrab(struct block_device *bdev); | |
2161 | extern void bd_set_size(struct block_device *, loff_t size); | |
2162 | extern void bd_forget(struct inode *inode); | |
2163 | extern void bdput(struct block_device *); | |
2164 | extern void invalidate_bdev(struct block_device *); | |
2165 | extern void iterate_bdevs(void (*)(struct block_device *, void *), void *); | |
2166 | extern int sync_blockdev(struct block_device *bdev); | |
2167 | extern void kill_bdev(struct block_device *); | |
2168 | extern struct super_block *freeze_bdev(struct block_device *); | |
2169 | extern void emergency_thaw_all(void); | |
2170 | extern int thaw_bdev(struct block_device *bdev, struct super_block *sb); | |
2171 | extern int fsync_bdev(struct block_device *); | |
2172 | extern int sb_is_blkdev_sb(struct super_block *sb); | |
2173 | #else | |
2174 | static inline void bd_forget(struct inode *inode) {} | |
2175 | static inline int sync_blockdev(struct block_device *bdev) { return 0; } | |
2176 | static inline void kill_bdev(struct block_device *bdev) {} | |
2177 | static inline void invalidate_bdev(struct block_device *bdev) {} | |
2178 | ||
2179 | static inline struct super_block *freeze_bdev(struct block_device *sb) | |
2180 | { | |
2181 | return NULL; | |
2182 | } | |
2183 | ||
2184 | static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb) | |
2185 | { | |
2186 | return 0; | |
2187 | } | |
2188 | ||
2189 | static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg) | |
2190 | { | |
2191 | } | |
2192 | ||
2193 | static inline int sb_is_blkdev_sb(struct super_block *sb) | |
2194 | { | |
2195 | return 0; | |
2196 | } | |
2197 | #endif | |
2198 | extern int sync_filesystem(struct super_block *); | |
2199 | extern const struct file_operations def_blk_fops; | |
2200 | extern const struct file_operations def_chr_fops; | |
2201 | #ifdef CONFIG_BLOCK | |
2202 | extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); | |
2203 | extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); | |
2204 | extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); | |
2205 | extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder); | |
2206 | extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode, | |
2207 | void *holder); | |
2208 | extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, | |
2209 | void *holder); | |
2210 | extern void blkdev_put(struct block_device *bdev, fmode_t mode); | |
2211 | #ifdef CONFIG_SYSFS | |
2212 | extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk); | |
2213 | extern void bd_unlink_disk_holder(struct block_device *bdev, | |
2214 | struct gendisk *disk); | |
2215 | #else | |
2216 | static inline int bd_link_disk_holder(struct block_device *bdev, | |
2217 | struct gendisk *disk) | |
2218 | { | |
2219 | return 0; | |
2220 | } | |
2221 | static inline void bd_unlink_disk_holder(struct block_device *bdev, | |
2222 | struct gendisk *disk) | |
2223 | { | |
2224 | } | |
2225 | #endif | |
2226 | #endif | |
2227 | ||
2228 | /* fs/char_dev.c */ | |
2229 | #define CHRDEV_MAJOR_HASH_SIZE 255 | |
2230 | extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); | |
2231 | extern int register_chrdev_region(dev_t, unsigned, const char *); | |
2232 | extern int __register_chrdev(unsigned int major, unsigned int baseminor, | |
2233 | unsigned int count, const char *name, | |
2234 | const struct file_operations *fops); | |
2235 | extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, | |
2236 | unsigned int count, const char *name); | |
2237 | extern void unregister_chrdev_region(dev_t, unsigned); | |
2238 | extern void chrdev_show(struct seq_file *,off_t); | |
2239 | ||
2240 | static inline int register_chrdev(unsigned int major, const char *name, | |
2241 | const struct file_operations *fops) | |
2242 | { | |
2243 | return __register_chrdev(major, 0, 256, name, fops); | |
2244 | } | |
2245 | ||
2246 | static inline void unregister_chrdev(unsigned int major, const char *name) | |
2247 | { | |
2248 | __unregister_chrdev(major, 0, 256, name); | |
2249 | } | |
2250 | ||
2251 | /* fs/block_dev.c */ | |
2252 | #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ | |
2253 | #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ | |
2254 | ||
2255 | #ifdef CONFIG_BLOCK | |
2256 | #define BLKDEV_MAJOR_HASH_SIZE 255 | |
2257 | extern const char *__bdevname(dev_t, char *buffer); | |
2258 | extern const char *bdevname(struct block_device *bdev, char *buffer); | |
2259 | extern struct block_device *lookup_bdev(const char *); | |
2260 | extern void blkdev_show(struct seq_file *,off_t); | |
2261 | ||
2262 | #else | |
2263 | #define BLKDEV_MAJOR_HASH_SIZE 0 | |
2264 | #endif | |
2265 | ||
2266 | extern void init_special_inode(struct inode *, umode_t, dev_t); | |
2267 | ||
2268 | /* Invalid inode operations -- fs/bad_inode.c */ | |
2269 | extern void make_bad_inode(struct inode *); | |
2270 | extern int is_bad_inode(struct inode *); | |
2271 | ||
2272 | #ifdef CONFIG_BLOCK | |
2273 | /* | |
2274 | * return READ, READA, or WRITE | |
2275 | */ | |
2276 | #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK)) | |
2277 | ||
2278 | /* | |
2279 | * return data direction, READ or WRITE | |
2280 | */ | |
2281 | #define bio_data_dir(bio) ((bio)->bi_rw & 1) | |
2282 | ||
2283 | extern void check_disk_size_change(struct gendisk *disk, | |
2284 | struct block_device *bdev); | |
2285 | extern int revalidate_disk(struct gendisk *); | |
2286 | extern int check_disk_change(struct block_device *); | |
2287 | extern int __invalidate_device(struct block_device *, bool); | |
2288 | extern int invalidate_partition(struct gendisk *, int); | |
2289 | #endif | |
2290 | unsigned long invalidate_mapping_pages(struct address_space *mapping, | |
2291 | pgoff_t start, pgoff_t end); | |
2292 | ||
2293 | static inline void invalidate_remote_inode(struct inode *inode) | |
2294 | { | |
2295 | if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || | |
2296 | S_ISLNK(inode->i_mode)) | |
2297 | invalidate_mapping_pages(inode->i_mapping, 0, -1); | |
2298 | } | |
2299 | extern int invalidate_inode_pages2(struct address_space *mapping); | |
2300 | extern int invalidate_inode_pages2_range(struct address_space *mapping, | |
2301 | pgoff_t start, pgoff_t end); | |
2302 | extern int write_inode_now(struct inode *, int); | |
2303 | extern int filemap_fdatawrite(struct address_space *); | |
2304 | extern int filemap_flush(struct address_space *); | |
2305 | extern int filemap_fdatawait(struct address_space *); | |
2306 | extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, | |
2307 | loff_t lend); | |
2308 | extern int filemap_write_and_wait(struct address_space *mapping); | |
2309 | extern int filemap_write_and_wait_range(struct address_space *mapping, | |
2310 | loff_t lstart, loff_t lend); | |
2311 | extern int __filemap_fdatawrite_range(struct address_space *mapping, | |
2312 | loff_t start, loff_t end, int sync_mode); | |
2313 | extern int filemap_fdatawrite_range(struct address_space *mapping, | |
2314 | loff_t start, loff_t end); | |
2315 | ||
2316 | extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, | |
2317 | int datasync); | |
2318 | extern int vfs_fsync(struct file *file, int datasync); | |
2319 | static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count) | |
2320 | { | |
2321 | if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host)) | |
2322 | return 0; | |
2323 | return vfs_fsync_range(file, pos, pos + count - 1, | |
2324 | (file->f_flags & __O_SYNC) ? 0 : 1); | |
2325 | } | |
2326 | extern void emergency_sync(void); | |
2327 | extern void emergency_remount(void); | |
2328 | #ifdef CONFIG_BLOCK | |
2329 | extern sector_t bmap(struct inode *, sector_t); | |
2330 | #endif | |
2331 | extern int notify_change(struct dentry *, struct iattr *, struct inode **); | |
2332 | extern int inode_permission(struct inode *, int); | |
2333 | extern int __inode_permission(struct inode *, int); | |
2334 | extern int generic_permission(struct inode *, int); | |
2335 | extern int __check_sticky(struct inode *dir, struct inode *inode); | |
2336 | ||
2337 | static inline bool execute_ok(struct inode *inode) | |
2338 | { | |
2339 | return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); | |
2340 | } | |
2341 | ||
2342 | static inline void file_start_write(struct file *file) | |
2343 | { | |
2344 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2345 | return; | |
2346 | __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true); | |
2347 | } | |
2348 | ||
2349 | static inline bool file_start_write_trylock(struct file *file) | |
2350 | { | |
2351 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2352 | return true; | |
2353 | return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false); | |
2354 | } | |
2355 | ||
2356 | static inline void file_end_write(struct file *file) | |
2357 | { | |
2358 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2359 | return; | |
2360 | __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE); | |
2361 | } | |
2362 | ||
2363 | /* | |
2364 | * get_write_access() gets write permission for a file. | |
2365 | * put_write_access() releases this write permission. | |
2366 | * This is used for regular files. | |
2367 | * We cannot support write (and maybe mmap read-write shared) accesses and | |
2368 | * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode | |
2369 | * can have the following values: | |
2370 | * 0: no writers, no VM_DENYWRITE mappings | |
2371 | * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist | |
2372 | * > 0: (i_writecount) users are writing to the file. | |
2373 | * | |
2374 | * Normally we operate on that counter with atomic_{inc,dec} and it's safe | |
2375 | * except for the cases where we don't hold i_writecount yet. Then we need to | |
2376 | * use {get,deny}_write_access() - these functions check the sign and refuse | |
2377 | * to do the change if sign is wrong. | |
2378 | */ | |
2379 | static inline int get_write_access(struct inode *inode) | |
2380 | { | |
2381 | return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; | |
2382 | } | |
2383 | static inline int deny_write_access(struct file *file) | |
2384 | { | |
2385 | struct inode *inode = file_inode(file); | |
2386 | return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; | |
2387 | } | |
2388 | static inline void put_write_access(struct inode * inode) | |
2389 | { | |
2390 | atomic_dec(&inode->i_writecount); | |
2391 | } | |
2392 | static inline void allow_write_access(struct file *file) | |
2393 | { | |
2394 | if (file) | |
2395 | atomic_inc(&file_inode(file)->i_writecount); | |
2396 | } | |
2397 | static inline bool inode_is_open_for_write(const struct inode *inode) | |
2398 | { | |
2399 | return atomic_read(&inode->i_writecount) > 0; | |
2400 | } | |
2401 | ||
2402 | #ifdef CONFIG_IMA | |
2403 | static inline void i_readcount_dec(struct inode *inode) | |
2404 | { | |
2405 | BUG_ON(!atomic_read(&inode->i_readcount)); | |
2406 | atomic_dec(&inode->i_readcount); | |
2407 | } | |
2408 | static inline void i_readcount_inc(struct inode *inode) | |
2409 | { | |
2410 | atomic_inc(&inode->i_readcount); | |
2411 | } | |
2412 | #else | |
2413 | static inline void i_readcount_dec(struct inode *inode) | |
2414 | { | |
2415 | return; | |
2416 | } | |
2417 | static inline void i_readcount_inc(struct inode *inode) | |
2418 | { | |
2419 | return; | |
2420 | } | |
2421 | #endif | |
2422 | extern int do_pipe_flags(int *, int); | |
2423 | ||
2424 | extern int kernel_read(struct file *, loff_t, char *, unsigned long); | |
2425 | extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t); | |
2426 | extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *); | |
2427 | extern struct file * open_exec(const char *); | |
2428 | ||
2429 | /* fs/dcache.c -- generic fs support functions */ | |
2430 | extern int is_subdir(struct dentry *, struct dentry *); | |
2431 | extern int path_is_under(struct path *, struct path *); | |
2432 | ||
2433 | #include <linux/err.h> | |
2434 | ||
2435 | /* needed for stackable file system support */ | |
2436 | extern loff_t default_llseek(struct file *file, loff_t offset, int whence); | |
2437 | ||
2438 | extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); | |
2439 | ||
2440 | extern int inode_init_always(struct super_block *, struct inode *); | |
2441 | extern void inode_init_once(struct inode *); | |
2442 | extern void address_space_init_once(struct address_space *mapping); | |
2443 | extern struct inode * igrab(struct inode *); | |
2444 | extern ino_t iunique(struct super_block *, ino_t); | |
2445 | extern int inode_needs_sync(struct inode *inode); | |
2446 | extern int generic_delete_inode(struct inode *inode); | |
2447 | static inline int generic_drop_inode(struct inode *inode) | |
2448 | { | |
2449 | return !inode->i_nlink || inode_unhashed(inode); | |
2450 | } | |
2451 | ||
2452 | extern struct inode *ilookup5_nowait(struct super_block *sb, | |
2453 | unsigned long hashval, int (*test)(struct inode *, void *), | |
2454 | void *data); | |
2455 | extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, | |
2456 | int (*test)(struct inode *, void *), void *data); | |
2457 | extern struct inode *ilookup(struct super_block *sb, unsigned long ino); | |
2458 | ||
2459 | extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); | |
2460 | extern struct inode * iget_locked(struct super_block *, unsigned long); | |
2461 | extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); | |
2462 | extern int insert_inode_locked(struct inode *); | |
2463 | #ifdef CONFIG_DEBUG_LOCK_ALLOC | |
2464 | extern void lockdep_annotate_inode_mutex_key(struct inode *inode); | |
2465 | #else | |
2466 | static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; | |
2467 | #endif | |
2468 | extern void unlock_new_inode(struct inode *); | |
2469 | extern unsigned int get_next_ino(void); | |
2470 | ||
2471 | extern void __iget(struct inode * inode); | |
2472 | extern void iget_failed(struct inode *); | |
2473 | extern void clear_inode(struct inode *); | |
2474 | extern void __destroy_inode(struct inode *); | |
2475 | extern struct inode *new_inode_pseudo(struct super_block *sb); | |
2476 | extern struct inode *new_inode(struct super_block *sb); | |
2477 | extern void free_inode_nonrcu(struct inode *inode); | |
2478 | extern int should_remove_suid(struct dentry *); | |
2479 | extern int file_remove_suid(struct file *); | |
2480 | ||
2481 | extern void __insert_inode_hash(struct inode *, unsigned long hashval); | |
2482 | static inline void insert_inode_hash(struct inode *inode) | |
2483 | { | |
2484 | __insert_inode_hash(inode, inode->i_ino); | |
2485 | } | |
2486 | ||
2487 | extern void __remove_inode_hash(struct inode *); | |
2488 | static inline void remove_inode_hash(struct inode *inode) | |
2489 | { | |
2490 | if (!inode_unhashed(inode)) | |
2491 | __remove_inode_hash(inode); | |
2492 | } | |
2493 | ||
2494 | extern void inode_sb_list_add(struct inode *inode); | |
2495 | ||
2496 | #ifdef CONFIG_BLOCK | |
2497 | extern void submit_bio(int, struct bio *); | |
2498 | extern int bdev_read_only(struct block_device *); | |
2499 | #endif | |
2500 | extern int set_blocksize(struct block_device *, int); | |
2501 | extern int sb_set_blocksize(struct super_block *, int); | |
2502 | extern int sb_min_blocksize(struct super_block *, int); | |
2503 | ||
2504 | extern int generic_file_mmap(struct file *, struct vm_area_struct *); | |
2505 | extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); | |
2506 | extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr, | |
2507 | unsigned long size, pgoff_t pgoff); | |
2508 | int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk); | |
2509 | extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); | |
2510 | extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); | |
2511 | extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); | |
2512 | extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t); | |
2513 | extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t); | |
2514 | extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos); | |
2515 | extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos); | |
2516 | extern ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos); | |
2517 | extern ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos); | |
2518 | ||
2519 | /* fs/block_dev.c */ | |
2520 | extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to); | |
2521 | extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
2522 | extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end, | |
2523 | int datasync); | |
2524 | extern void block_sync_page(struct page *page); | |
2525 | ||
2526 | /* fs/splice.c */ | |
2527 | extern ssize_t generic_file_splice_read(struct file *, loff_t *, | |
2528 | struct pipe_inode_info *, size_t, unsigned int); | |
2529 | extern ssize_t default_file_splice_read(struct file *, loff_t *, | |
2530 | struct pipe_inode_info *, size_t, unsigned int); | |
2531 | extern ssize_t iter_file_splice_write(struct pipe_inode_info *, | |
2532 | struct file *, loff_t *, size_t, unsigned int); | |
2533 | extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, | |
2534 | struct file *out, loff_t *, size_t len, unsigned int flags); | |
2535 | extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, | |
2536 | loff_t *opos, size_t len, unsigned int flags); | |
2537 | ||
2538 | ||
2539 | extern void | |
2540 | file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); | |
2541 | extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); | |
2542 | extern loff_t no_llseek(struct file *file, loff_t offset, int whence); | |
2543 | extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); | |
2544 | extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); | |
2545 | extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, | |
2546 | int whence, loff_t maxsize, loff_t eof); | |
2547 | extern loff_t fixed_size_llseek(struct file *file, loff_t offset, | |
2548 | int whence, loff_t size); | |
2549 | extern int generic_file_open(struct inode * inode, struct file * filp); | |
2550 | extern int nonseekable_open(struct inode * inode, struct file * filp); | |
2551 | ||
2552 | #ifdef CONFIG_FS_XIP | |
2553 | extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len, | |
2554 | loff_t *ppos); | |
2555 | extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma); | |
2556 | extern ssize_t xip_file_write(struct file *filp, const char __user *buf, | |
2557 | size_t len, loff_t *ppos); | |
2558 | extern int xip_truncate_page(struct address_space *mapping, loff_t from); | |
2559 | #else | |
2560 | static inline int xip_truncate_page(struct address_space *mapping, loff_t from) | |
2561 | { | |
2562 | return 0; | |
2563 | } | |
2564 | #endif | |
2565 | ||
2566 | #ifdef CONFIG_BLOCK | |
2567 | typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode, | |
2568 | loff_t file_offset); | |
2569 | ||
2570 | enum { | |
2571 | /* need locking between buffered and direct access */ | |
2572 | DIO_LOCKING = 0x01, | |
2573 | ||
2574 | /* filesystem does not support filling holes */ | |
2575 | DIO_SKIP_HOLES = 0x02, | |
2576 | ||
2577 | /* filesystem can handle aio writes beyond i_size */ | |
2578 | DIO_ASYNC_EXTEND = 0x04, | |
2579 | }; | |
2580 | ||
2581 | void dio_end_io(struct bio *bio, int error); | |
2582 | ||
2583 | ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode, | |
2584 | struct block_device *bdev, struct iov_iter *iter, loff_t offset, | |
2585 | get_block_t get_block, dio_iodone_t end_io, | |
2586 | dio_submit_t submit_io, int flags); | |
2587 | ||
2588 | static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb, | |
2589 | struct inode *inode, struct iov_iter *iter, loff_t offset, | |
2590 | get_block_t get_block) | |
2591 | { | |
2592 | return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iter, | |
2593 | offset, get_block, NULL, NULL, | |
2594 | DIO_LOCKING | DIO_SKIP_HOLES); | |
2595 | } | |
2596 | #endif | |
2597 | ||
2598 | void inode_dio_wait(struct inode *inode); | |
2599 | void inode_dio_done(struct inode *inode); | |
2600 | ||
2601 | extern void inode_set_flags(struct inode *inode, unsigned int flags, | |
2602 | unsigned int mask); | |
2603 | ||
2604 | extern const struct file_operations generic_ro_fops; | |
2605 | ||
2606 | #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) | |
2607 | ||
2608 | extern int readlink_copy(char __user *, int, const char *); | |
2609 | extern int page_readlink(struct dentry *, char __user *, int); | |
2610 | extern void *page_follow_link_light(struct dentry *, struct nameidata *); | |
2611 | extern void page_put_link(struct dentry *, struct nameidata *, void *); | |
2612 | extern int __page_symlink(struct inode *inode, const char *symname, int len, | |
2613 | int nofs); | |
2614 | extern int page_symlink(struct inode *inode, const char *symname, int len); | |
2615 | extern const struct inode_operations page_symlink_inode_operations; | |
2616 | extern void kfree_put_link(struct dentry *, struct nameidata *, void *); | |
2617 | extern int generic_readlink(struct dentry *, char __user *, int); | |
2618 | extern void generic_fillattr(struct inode *, struct kstat *); | |
2619 | int vfs_getattr_nosec(struct path *path, struct kstat *stat); | |
2620 | extern int vfs_getattr(struct path *, struct kstat *); | |
2621 | void __inode_add_bytes(struct inode *inode, loff_t bytes); | |
2622 | void inode_add_bytes(struct inode *inode, loff_t bytes); | |
2623 | void __inode_sub_bytes(struct inode *inode, loff_t bytes); | |
2624 | void inode_sub_bytes(struct inode *inode, loff_t bytes); | |
2625 | loff_t inode_get_bytes(struct inode *inode); | |
2626 | void inode_set_bytes(struct inode *inode, loff_t bytes); | |
2627 | ||
2628 | extern int vfs_readdir(struct file *, filldir_t, void *); | |
2629 | extern int iterate_dir(struct file *, struct dir_context *); | |
2630 | ||
2631 | extern int vfs_stat(const char __user *, struct kstat *); | |
2632 | extern int vfs_lstat(const char __user *, struct kstat *); | |
2633 | extern int vfs_fstat(unsigned int, struct kstat *); | |
2634 | extern int vfs_fstatat(int , const char __user *, struct kstat *, int); | |
2635 | ||
2636 | extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd, | |
2637 | unsigned long arg); | |
2638 | extern int __generic_block_fiemap(struct inode *inode, | |
2639 | struct fiemap_extent_info *fieinfo, | |
2640 | loff_t start, loff_t len, | |
2641 | get_block_t *get_block); | |
2642 | extern int generic_block_fiemap(struct inode *inode, | |
2643 | struct fiemap_extent_info *fieinfo, u64 start, | |
2644 | u64 len, get_block_t *get_block); | |
2645 | ||
2646 | extern void get_filesystem(struct file_system_type *fs); | |
2647 | extern void put_filesystem(struct file_system_type *fs); | |
2648 | extern struct file_system_type *get_fs_type(const char *name); | |
2649 | extern struct super_block *get_super(struct block_device *); | |
2650 | extern struct super_block *get_super_thawed(struct block_device *); | |
2651 | extern struct super_block *get_active_super(struct block_device *bdev); | |
2652 | extern void drop_super(struct super_block *sb); | |
2653 | extern void iterate_supers(void (*)(struct super_block *, void *), void *); | |
2654 | extern void iterate_supers_type(struct file_system_type *, | |
2655 | void (*)(struct super_block *, void *), void *); | |
2656 | ||
2657 | extern int dcache_dir_open(struct inode *, struct file *); | |
2658 | extern int dcache_dir_close(struct inode *, struct file *); | |
2659 | extern loff_t dcache_dir_lseek(struct file *, loff_t, int); | |
2660 | extern int dcache_readdir(struct file *, struct dir_context *); | |
2661 | extern int simple_setattr(struct dentry *, struct iattr *); | |
2662 | extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *); | |
2663 | extern int simple_statfs(struct dentry *, struct kstatfs *); | |
2664 | extern int simple_open(struct inode *inode, struct file *file); | |
2665 | extern int simple_link(struct dentry *, struct inode *, struct dentry *); | |
2666 | extern int simple_unlink(struct inode *, struct dentry *); | |
2667 | extern int simple_rmdir(struct inode *, struct dentry *); | |
2668 | extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *); | |
2669 | extern int noop_fsync(struct file *, loff_t, loff_t, int); | |
2670 | extern int simple_empty(struct dentry *); | |
2671 | extern int simple_readpage(struct file *file, struct page *page); | |
2672 | extern int simple_write_begin(struct file *file, struct address_space *mapping, | |
2673 | loff_t pos, unsigned len, unsigned flags, | |
2674 | struct page **pagep, void **fsdata); | |
2675 | extern int simple_write_end(struct file *file, struct address_space *mapping, | |
2676 | loff_t pos, unsigned len, unsigned copied, | |
2677 | struct page *page, void *fsdata); | |
2678 | extern int always_delete_dentry(const struct dentry *); | |
2679 | extern struct inode *alloc_anon_inode(struct super_block *); | |
2680 | extern int simple_nosetlease(struct file *, long, struct file_lock **, void **); | |
2681 | extern const struct dentry_operations simple_dentry_operations; | |
2682 | ||
2683 | extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); | |
2684 | extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); | |
2685 | extern const struct file_operations simple_dir_operations; | |
2686 | extern const struct inode_operations simple_dir_inode_operations; | |
2687 | struct tree_descr { char *name; const struct file_operations *ops; int mode; }; | |
2688 | struct dentry *d_alloc_name(struct dentry *, const char *); | |
2689 | extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *); | |
2690 | extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); | |
2691 | extern void simple_release_fs(struct vfsmount **mount, int *count); | |
2692 | ||
2693 | extern ssize_t simple_read_from_buffer(void __user *to, size_t count, | |
2694 | loff_t *ppos, const void *from, size_t available); | |
2695 | extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, | |
2696 | const void __user *from, size_t count); | |
2697 | ||
2698 | extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); | |
2699 | extern int generic_file_fsync(struct file *, loff_t, loff_t, int); | |
2700 | ||
2701 | extern int generic_check_addressable(unsigned, u64); | |
2702 | ||
2703 | #ifdef CONFIG_MIGRATION | |
2704 | extern int buffer_migrate_page(struct address_space *, | |
2705 | struct page *, struct page *, | |
2706 | enum migrate_mode); | |
2707 | #else | |
2708 | #define buffer_migrate_page NULL | |
2709 | #endif | |
2710 | ||
2711 | extern int inode_change_ok(const struct inode *, struct iattr *); | |
2712 | extern int inode_newsize_ok(const struct inode *, loff_t offset); | |
2713 | extern void setattr_copy(struct inode *inode, const struct iattr *attr); | |
2714 | ||
2715 | extern int file_update_time(struct file *file); | |
2716 | ||
2717 | extern int generic_show_options(struct seq_file *m, struct dentry *root); | |
2718 | extern void save_mount_options(struct super_block *sb, char *options); | |
2719 | extern void replace_mount_options(struct super_block *sb, char *options); | |
2720 | ||
2721 | static inline ino_t parent_ino(struct dentry *dentry) | |
2722 | { | |
2723 | ino_t res; | |
2724 | ||
2725 | /* | |
2726 | * Don't strictly need d_lock here? If the parent ino could change | |
2727 | * then surely we'd have a deeper race in the caller? | |
2728 | */ | |
2729 | spin_lock(&dentry->d_lock); | |
2730 | res = dentry->d_parent->d_inode->i_ino; | |
2731 | spin_unlock(&dentry->d_lock); | |
2732 | return res; | |
2733 | } | |
2734 | ||
2735 | /* Transaction based IO helpers */ | |
2736 | ||
2737 | /* | |
2738 | * An argresp is stored in an allocated page and holds the | |
2739 | * size of the argument or response, along with its content | |
2740 | */ | |
2741 | struct simple_transaction_argresp { | |
2742 | ssize_t size; | |
2743 | char data[0]; | |
2744 | }; | |
2745 | ||
2746 | #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) | |
2747 | ||
2748 | char *simple_transaction_get(struct file *file, const char __user *buf, | |
2749 | size_t size); | |
2750 | ssize_t simple_transaction_read(struct file *file, char __user *buf, | |
2751 | size_t size, loff_t *pos); | |
2752 | int simple_transaction_release(struct inode *inode, struct file *file); | |
2753 | ||
2754 | void simple_transaction_set(struct file *file, size_t n); | |
2755 | ||
2756 | /* | |
2757 | * simple attribute files | |
2758 | * | |
2759 | * These attributes behave similar to those in sysfs: | |
2760 | * | |
2761 | * Writing to an attribute immediately sets a value, an open file can be | |
2762 | * written to multiple times. | |
2763 | * | |
2764 | * Reading from an attribute creates a buffer from the value that might get | |
2765 | * read with multiple read calls. When the attribute has been read | |
2766 | * completely, no further read calls are possible until the file is opened | |
2767 | * again. | |
2768 | * | |
2769 | * All attributes contain a text representation of a numeric value | |
2770 | * that are accessed with the get() and set() functions. | |
2771 | */ | |
2772 | #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ | |
2773 | static int __fops ## _open(struct inode *inode, struct file *file) \ | |
2774 | { \ | |
2775 | __simple_attr_check_format(__fmt, 0ull); \ | |
2776 | return simple_attr_open(inode, file, __get, __set, __fmt); \ | |
2777 | } \ | |
2778 | static const struct file_operations __fops = { \ | |
2779 | .owner = THIS_MODULE, \ | |
2780 | .open = __fops ## _open, \ | |
2781 | .release = simple_attr_release, \ | |
2782 | .read = simple_attr_read, \ | |
2783 | .write = simple_attr_write, \ | |
2784 | .llseek = generic_file_llseek, \ | |
2785 | } | |
2786 | ||
2787 | static inline __printf(1, 2) | |
2788 | void __simple_attr_check_format(const char *fmt, ...) | |
2789 | { | |
2790 | /* don't do anything, just let the compiler check the arguments; */ | |
2791 | } | |
2792 | ||
2793 | int simple_attr_open(struct inode *inode, struct file *file, | |
2794 | int (*get)(void *, u64 *), int (*set)(void *, u64), | |
2795 | const char *fmt); | |
2796 | int simple_attr_release(struct inode *inode, struct file *file); | |
2797 | ssize_t simple_attr_read(struct file *file, char __user *buf, | |
2798 | size_t len, loff_t *ppos); | |
2799 | ssize_t simple_attr_write(struct file *file, const char __user *buf, | |
2800 | size_t len, loff_t *ppos); | |
2801 | ||
2802 | struct ctl_table; | |
2803 | int proc_nr_files(struct ctl_table *table, int write, | |
2804 | void __user *buffer, size_t *lenp, loff_t *ppos); | |
2805 | int proc_nr_dentry(struct ctl_table *table, int write, | |
2806 | void __user *buffer, size_t *lenp, loff_t *ppos); | |
2807 | int proc_nr_inodes(struct ctl_table *table, int write, | |
2808 | void __user *buffer, size_t *lenp, loff_t *ppos); | |
2809 | int __init get_filesystem_list(char *buf); | |
2810 | ||
2811 | #define __FMODE_EXEC ((__force int) FMODE_EXEC) | |
2812 | #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) | |
2813 | ||
2814 | #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) | |
2815 | #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ | |
2816 | (flag & __FMODE_NONOTIFY))) | |
2817 | ||
2818 | static inline int is_sxid(umode_t mode) | |
2819 | { | |
2820 | return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP)); | |
2821 | } | |
2822 | ||
2823 | static inline int check_sticky(struct inode *dir, struct inode *inode) | |
2824 | { | |
2825 | if (!(dir->i_mode & S_ISVTX)) | |
2826 | return 0; | |
2827 | ||
2828 | return __check_sticky(dir, inode); | |
2829 | } | |
2830 | ||
2831 | static inline void inode_has_no_xattr(struct inode *inode) | |
2832 | { | |
2833 | if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC)) | |
2834 | inode->i_flags |= S_NOSEC; | |
2835 | } | |
2836 | ||
2837 | static inline bool is_root_inode(struct inode *inode) | |
2838 | { | |
2839 | return inode == inode->i_sb->s_root->d_inode; | |
2840 | } | |
2841 | ||
2842 | static inline bool dir_emit(struct dir_context *ctx, | |
2843 | const char *name, int namelen, | |
2844 | u64 ino, unsigned type) | |
2845 | { | |
2846 | return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0; | |
2847 | } | |
2848 | static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) | |
2849 | { | |
2850 | return ctx->actor(ctx, ".", 1, ctx->pos, | |
2851 | file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0; | |
2852 | } | |
2853 | static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) | |
2854 | { | |
2855 | return ctx->actor(ctx, "..", 2, ctx->pos, | |
2856 | parent_ino(file->f_path.dentry), DT_DIR) == 0; | |
2857 | } | |
2858 | static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) | |
2859 | { | |
2860 | if (ctx->pos == 0) { | |
2861 | if (!dir_emit_dot(file, ctx)) | |
2862 | return false; | |
2863 | ctx->pos = 1; | |
2864 | } | |
2865 | if (ctx->pos == 1) { | |
2866 | if (!dir_emit_dotdot(file, ctx)) | |
2867 | return false; | |
2868 | ctx->pos = 2; | |
2869 | } | |
2870 | return true; | |
2871 | } | |
2872 | static inline bool dir_relax(struct inode *inode) | |
2873 | { | |
2874 | mutex_unlock(&inode->i_mutex); | |
2875 | mutex_lock(&inode->i_mutex); | |
2876 | return !IS_DEADDIR(inode); | |
2877 | } | |
2878 | ||
2879 | #endif /* _LINUX_FS_H */ |