Commit | Line | Data |
---|---|---|
de57606c SW |
1 | #ifndef _FS_CEPH_SUPER_H |
2 | #define _FS_CEPH_SUPER_H | |
3 | ||
3d14c5d2 | 4 | #include <linux/ceph/ceph_debug.h> |
de57606c SW |
5 | |
6 | #include <asm/unaligned.h> | |
7 | #include <linux/backing-dev.h> | |
8 | #include <linux/completion.h> | |
9 | #include <linux/exportfs.h> | |
10 | #include <linux/fs.h> | |
11 | #include <linux/mempool.h> | |
12 | #include <linux/pagemap.h> | |
13 | #include <linux/wait.h> | |
f1a3d572 | 14 | #include <linux/writeback.h> |
5a0e3ad6 | 15 | #include <linux/slab.h> |
c969d9bf | 16 | #include <linux/posix_acl.h> |
805692d0 | 17 | #include <linux/refcount.h> |
de57606c | 18 | |
3d14c5d2 | 19 | #include <linux/ceph/libceph.h> |
de57606c | 20 | |
99ccbd22 MT |
21 | #ifdef CONFIG_CEPH_FSCACHE |
22 | #include <linux/fscache.h> | |
23 | #endif | |
24 | ||
de57606c SW |
25 | /* f_type in struct statfs */ |
26 | #define CEPH_SUPER_MAGIC 0x00c36400 | |
27 | ||
28 | /* large granularity for statfs utilization stats to facilitate | |
29 | * large volume sizes on 32-bit machines. */ | |
92a49fb0 | 30 | #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ |
de57606c SW |
31 | #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) |
32 | ||
3d14c5d2 YS |
33 | #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ |
34 | #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ | |
35 | #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ | |
ad1fee96 | 36 | #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ |
a40dc6cc | 37 | #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ |
99ccbd22 | 38 | #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ |
10183a69 | 39 | #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ |
e9e427f0 | 40 | #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ |
6a259382 | 41 | |
133e9156 | 42 | #define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE |
de57606c | 43 | |
3d14c5d2 YS |
44 | #define ceph_set_mount_opt(fsc, opt) \ |
45 | (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; | |
46 | #define ceph_test_mount_opt(fsc, opt) \ | |
47 | (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) | |
de57606c | 48 | |
aa187926 YZ |
49 | /* max size of osd read request, limited by libceph */ |
50 | #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN | |
95cca2b4 YZ |
51 | /* osd has a configurable limitaion of max write size. |
52 | * CEPH_MSG_MAX_DATA_LEN should be small enough. */ | |
53 | #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN | |
7c94ba27 | 54 | #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ |
3d14c5d2 YS |
55 | #define CEPH_MAX_READDIR_DEFAULT 1024 |
56 | #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) | |
57 | #define CEPH_SNAPDIRNAME_DEFAULT ".snap" | |
de57606c | 58 | |
3d14c5d2 | 59 | struct ceph_mount_options { |
6e19a16e | 60 | int flags; |
3d14c5d2 YS |
61 | int sb_flags; |
62 | ||
83817e35 SW |
63 | int wsize; /* max write size */ |
64 | int rsize; /* max read size */ | |
65 | int rasize; /* max readahead */ | |
6e19a16e SW |
66 | int congestion_kb; /* max writeback in flight */ |
67 | int caps_wanted_delay_min, caps_wanted_delay_max; | |
23804d91 SW |
68 | int max_readdir; /* max readdir result (entires) */ |
69 | int max_readdir_bytes; /* max readdir result (bytes) */ | |
de57606c | 70 | |
3d14c5d2 YS |
71 | /* |
72 | * everything above this point can be memcmp'd; everything below | |
73 | * is handled in compare_mount_options() | |
74 | */ | |
de57606c | 75 | |
3d14c5d2 | 76 | char *snapdir_name; /* default ".snap" */ |
430afbad | 77 | char *mds_namespace; /* default NULL */ |
3f384954 | 78 | char *server_path; /* default "/" */ |
1d8f8360 | 79 | char *fscache_uniq; /* default NULL */ |
3d14c5d2 | 80 | }; |
de57606c | 81 | |
3d14c5d2 | 82 | struct ceph_fs_client { |
de57606c SW |
83 | struct super_block *sb; |
84 | ||
3d14c5d2 YS |
85 | struct ceph_mount_options *mount_options; |
86 | struct ceph_client *client; | |
de57606c | 87 | |
3d14c5d2 | 88 | unsigned long mount_state; |
85ccce43 SW |
89 | int min_caps; /* min caps i added */ |
90 | ||
3d14c5d2 | 91 | struct ceph_mds_client *mdsc; |
de57606c SW |
92 | |
93 | /* writeback */ | |
94 | mempool_t *wb_pagevec_pool; | |
95 | struct workqueue_struct *wb_wq; | |
96 | struct workqueue_struct *pg_inv_wq; | |
97 | struct workqueue_struct *trunc_wq; | |
2baba250 | 98 | atomic_long_t writeback_count; |
de57606c | 99 | |
0743304d | 100 | #ifdef CONFIG_DEBUG_FS |
3d14c5d2 | 101 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
2baba250 | 102 | struct dentry *debugfs_congestion_kb; |
06edf046 | 103 | struct dentry *debugfs_bdi; |
3d14c5d2 | 104 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
14ed9703 | 105 | struct dentry *debugfs_mds_sessions; |
0743304d | 106 | #endif |
99ccbd22 MT |
107 | |
108 | #ifdef CONFIG_CEPH_FSCACHE | |
109 | struct fscache_cookie *fscache; | |
99ccbd22 | 110 | #endif |
de57606c SW |
111 | }; |
112 | ||
3d14c5d2 | 113 | |
de57606c SW |
114 | /* |
115 | * File i/o capability. This tracks shared state with the metadata | |
116 | * server that allows us to cache or writeback attributes or to read | |
117 | * and write data. For any given inode, we should have one or more | |
118 | * capabilities, one issued by each metadata server, and our | |
119 | * cumulative access is the OR of all issued capabilities. | |
120 | * | |
121 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds | |
122 | * session capability lists. | |
123 | */ | |
124 | struct ceph_cap { | |
125 | struct ceph_inode_info *ci; | |
126 | struct rb_node ci_node; /* per-ci cap tree */ | |
127 | struct ceph_mds_session *session; | |
128 | struct list_head session_caps; /* per-session caplist */ | |
de57606c | 129 | u64 cap_id; /* unique cap id (mds provided) */ |
745a8e3b YZ |
130 | union { |
131 | /* in-use caps */ | |
132 | struct { | |
133 | int issued; /* latest, from the mds */ | |
134 | int implemented; /* implemented superset of | |
135 | issued (for revocation) */ | |
136 | int mds, mds_wanted; | |
137 | }; | |
138 | /* caps to release */ | |
139 | struct { | |
140 | u64 cap_ino; | |
141 | int queue_release; | |
142 | }; | |
143 | }; | |
685f9a5d SW |
144 | u32 seq, issue_seq, mseq; |
145 | u32 cap_gen; /* active/stale cycle */ | |
de57606c SW |
146 | unsigned long last_used; |
147 | struct list_head caps_item; | |
148 | }; | |
149 | ||
150 | #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ | |
151 | #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ | |
152 | #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ | |
153 | ||
0e294387 YZ |
154 | struct ceph_cap_flush { |
155 | u64 tid; | |
156 | int caps; /* 0 means capsnap */ | |
c8799fc4 | 157 | bool wake; /* wake up flush waiters when finish ? */ |
0e294387 YZ |
158 | struct list_head g_list; // global |
159 | struct list_head i_list; // per inode | |
160 | }; | |
161 | ||
de57606c SW |
162 | /* |
163 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, | |
164 | * we first complete any in-process sync writes and writeback any dirty | |
165 | * data before flushing the snapped state (tracked here) back to the MDS. | |
166 | */ | |
167 | struct ceph_cap_snap { | |
805692d0 | 168 | refcount_t nref; |
0e294387 | 169 | struct list_head ci_item; |
de57606c | 170 | |
0e294387 YZ |
171 | struct ceph_cap_flush cap_flush; |
172 | ||
173 | u64 follows; | |
de57606c SW |
174 | int issued, dirty; |
175 | struct ceph_snap_context *context; | |
176 | ||
5706b27d | 177 | umode_t mode; |
05cb11c1 EB |
178 | kuid_t uid; |
179 | kgid_t gid; | |
de57606c | 180 | |
4a625be4 | 181 | struct ceph_buffer *xattr_blob; |
de57606c SW |
182 | u64 xattr_version; |
183 | ||
184 | u64 size; | |
185 | struct timespec mtime, atime, ctime; | |
186 | u64 time_warp_seq; | |
5f743e45 YZ |
187 | u64 truncate_size; |
188 | u32 truncate_seq; | |
de57606c SW |
189 | int writing; /* a sync write is still in progress */ |
190 | int dirty_pages; /* dirty pages awaiting writeback */ | |
e20d258d | 191 | bool inline_data; |
86056090 | 192 | bool need_flush; |
de57606c SW |
193 | }; |
194 | ||
195 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) | |
196 | { | |
805692d0 | 197 | if (refcount_dec_and_test(&capsnap->nref)) { |
4a625be4 SW |
198 | if (capsnap->xattr_blob) |
199 | ceph_buffer_put(capsnap->xattr_blob); | |
de57606c | 200 | kfree(capsnap); |
4a625be4 | 201 | } |
de57606c SW |
202 | } |
203 | ||
204 | /* | |
205 | * The frag tree describes how a directory is fragmented, potentially across | |
206 | * multiple metadata servers. It is also used to indicate points where | |
207 | * metadata authority is delegated, and whether/where metadata is replicated. | |
208 | * | |
209 | * A _leaf_ frag will be present in the i_fragtree IFF there is | |
210 | * delegation info. That is, if mds >= 0 || ndist > 0. | |
211 | */ | |
212 | #define CEPH_MAX_DIRFRAG_REP 4 | |
213 | ||
214 | struct ceph_inode_frag { | |
215 | struct rb_node node; | |
216 | ||
217 | /* fragtree state */ | |
218 | u32 frag; | |
219 | int split_by; /* i.e. 2^(split_by) children */ | |
220 | ||
221 | /* delegation and replication info */ | |
222 | int mds; /* -1 if same authority as parent */ | |
223 | int ndist; /* >0 if replicated */ | |
224 | int dist[CEPH_MAX_DIRFRAG_REP]; | |
225 | }; | |
226 | ||
227 | /* | |
228 | * We cache inode xattrs as an encoded blob until they are first used, | |
229 | * at which point we parse them into an rbtree. | |
230 | */ | |
231 | struct ceph_inode_xattr { | |
232 | struct rb_node node; | |
233 | ||
234 | const char *name; | |
235 | int name_len; | |
236 | const char *val; | |
237 | int val_len; | |
238 | int dirty; | |
239 | ||
240 | int should_free_name; | |
241 | int should_free_val; | |
242 | }; | |
243 | ||
3d14c5d2 YS |
244 | /* |
245 | * Ceph dentry state | |
246 | */ | |
247 | struct ceph_dentry_info { | |
248 | struct ceph_mds_session *lease_session; | |
249 | u32 lease_gen, lease_shared_gen; | |
250 | u32 lease_seq; | |
251 | unsigned long lease_renew_after, lease_renew_from; | |
252 | struct list_head lru; | |
253 | struct dentry *dentry; | |
9b16f03c | 254 | unsigned long time; |
3d14c5d2 YS |
255 | u64 offset; |
256 | }; | |
257 | ||
de57606c SW |
258 | struct ceph_inode_xattrs_info { |
259 | /* | |
260 | * (still encoded) xattr blob. we avoid the overhead of parsing | |
261 | * this until someone actually calls getxattr, etc. | |
262 | * | |
263 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == | |
264 | * NULL means we don't know. | |
265 | */ | |
266 | struct ceph_buffer *blob, *prealloc_blob; | |
267 | ||
268 | struct rb_root index; | |
269 | bool dirty; | |
270 | int count; | |
271 | int names_size; | |
272 | int vals_size; | |
273 | u64 version, index_version; | |
274 | }; | |
275 | ||
276 | /* | |
277 | * Ceph inode. | |
278 | */ | |
de57606c SW |
279 | struct ceph_inode_info { |
280 | struct ceph_vino i_vino; /* ceph ino + snap */ | |
281 | ||
be655596 SW |
282 | spinlock_t i_ceph_lock; |
283 | ||
de57606c | 284 | u64 i_version; |
31c542a1 | 285 | u64 i_inline_version; |
de57606c SW |
286 | u32 i_time_warp_seq; |
287 | ||
288 | unsigned i_ceph_flags; | |
fdd4e158 YZ |
289 | atomic64_t i_release_count; |
290 | atomic64_t i_ordered_count; | |
291 | atomic64_t i_complete_seq[2]; | |
de57606c | 292 | |
6c0f3af7 | 293 | struct ceph_dir_layout i_dir_layout; |
de57606c SW |
294 | struct ceph_file_layout i_layout; |
295 | char *i_symlink; | |
296 | ||
297 | /* for dirs */ | |
298 | struct timespec i_rctime; | |
299 | u64 i_rbytes, i_rfiles, i_rsubdirs; | |
300 | u64 i_files, i_subdirs; | |
de57606c SW |
301 | |
302 | struct rb_root i_fragtree; | |
1b1bc16d | 303 | int i_fragtree_nsplits; |
de57606c SW |
304 | struct mutex i_fragtree_mutex; |
305 | ||
306 | struct ceph_inode_xattrs_info i_xattrs; | |
307 | ||
be655596 | 308 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
de57606c SW |
309 | * s_mutex. */ |
310 | struct rb_root i_caps; /* cap list */ | |
311 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ | |
312 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ | |
313 | struct list_head i_dirty_item, i_flushing_item; | |
de57606c SW |
314 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
315 | * overlapping, pipelined cap flushes to the mds. we can probably | |
316 | * reduce the tid to 8 bits if we're concerned about inode size. */ | |
f66fd9f0 | 317 | struct ceph_cap_flush *i_prealloc_cap_flush; |
e4500b5e | 318 | struct list_head i_cap_flush_list; |
de57606c SW |
319 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
320 | unsigned long i_hold_caps_min; /* jiffies */ | |
321 | unsigned long i_hold_caps_max; /* jiffies */ | |
322 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ | |
de57606c SW |
323 | struct ceph_cap_reservation i_cap_migration_resv; |
324 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ | |
7d8cb26d SW |
325 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
326 | dirty|flushing caps */ | |
de57606c SW |
327 | unsigned i_snap_caps; /* cap bits for snapped files */ |
328 | ||
774a6a11 | 329 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
de57606c | 330 | |
b0d7c223 | 331 | struct mutex i_truncate_mutex; |
de57606c SW |
332 | u32 i_truncate_seq; /* last truncate to smaller size */ |
333 | u64 i_truncate_size; /* and the size we last truncated down to */ | |
334 | int i_truncate_pending; /* still need to call vmtruncate */ | |
335 | ||
336 | u64 i_max_size; /* max file size authorized by mds */ | |
337 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ | |
338 | u64 i_wanted_max_size; /* offset we'd like to write too */ | |
339 | u64 i_requested_max_size; /* max_size we've requested */ | |
340 | ||
341 | /* held references to caps */ | |
342 | int i_pin_ref; | |
d3d0720d | 343 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; |
de57606c SW |
344 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
345 | u32 i_shared_gen; /* increment each time we get FILE_SHARED */ | |
cd045cb4 | 346 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
de57606c SW |
347 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
348 | ||
de57606c | 349 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
68cd5b4b | 350 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
de57606c SW |
351 | spinlock_t i_unsafe_lock; |
352 | ||
353 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ | |
354 | int i_snap_realm_counter; /* snap realm (if caps) */ | |
355 | struct list_head i_snap_realm_item; | |
356 | struct list_head i_snap_flush_item; | |
357 | ||
358 | struct work_struct i_wb_work; /* writeback work */ | |
359 | struct work_struct i_pg_inv_work; /* page invalidation work */ | |
360 | ||
361 | struct work_struct i_vmtruncate_work; | |
362 | ||
99ccbd22 MT |
363 | #ifdef CONFIG_CEPH_FSCACHE |
364 | struct fscache_cookie *fscache; | |
f7f7e7a0 | 365 | u32 i_fscache_gen; |
99ccbd22 | 366 | #endif |
de57606c SW |
367 | struct inode vfs_inode; /* at end */ |
368 | }; | |
369 | ||
370 | static inline struct ceph_inode_info *ceph_inode(struct inode *inode) | |
371 | { | |
fbbccec9 | 372 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
de57606c SW |
373 | } |
374 | ||
ad1fee96 YS |
375 | static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) |
376 | { | |
377 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; | |
378 | } | |
379 | ||
380 | static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) | |
381 | { | |
382 | return (struct ceph_fs_client *)sb->s_fs_info; | |
383 | } | |
384 | ||
3d14c5d2 YS |
385 | static inline struct ceph_vino ceph_vino(struct inode *inode) |
386 | { | |
387 | return ceph_inode(inode)->i_vino; | |
388 | } | |
389 | ||
390 | /* | |
391 | * ino_t is <64 bits on many architectures, blech. | |
392 | * | |
ad1fee96 YS |
393 | * i_ino (kernel inode) st_ino (userspace) |
394 | * i386 32 32 | |
395 | * x86_64+ino32 64 32 | |
396 | * x86_64 64 64 | |
397 | */ | |
3310f754 | 398 | static inline u32 ceph_ino_to_ino32(__u64 vino) |
ad1fee96 | 399 | { |
3310f754 AO |
400 | u32 ino = vino & 0xffffffff; |
401 | ino ^= vino >> 32; | |
ad1fee96 | 402 | if (!ino) |
a661fc56 | 403 | ino = 2; |
ad1fee96 YS |
404 | return ino; |
405 | } | |
406 | ||
407 | /* | |
408 | * kernel i_ino value | |
3d14c5d2 YS |
409 | */ |
410 | static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) | |
411 | { | |
3d14c5d2 | 412 | #if BITS_PER_LONG == 32 |
3310f754 AO |
413 | return ceph_ino_to_ino32(vino.ino); |
414 | #else | |
415 | return (ino_t)vino.ino; | |
3d14c5d2 | 416 | #endif |
3d14c5d2 YS |
417 | } |
418 | ||
ad1fee96 YS |
419 | /* |
420 | * user-visible ino (stat, filldir) | |
421 | */ | |
422 | #if BITS_PER_LONG == 32 | |
423 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
424 | { | |
425 | return ino; | |
426 | } | |
427 | #else | |
428 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
429 | { | |
430 | if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) | |
431 | ino = ceph_ino_to_ino32(ino); | |
432 | return ino; | |
433 | } | |
434 | #endif | |
435 | ||
436 | ||
3d14c5d2 YS |
437 | /* for printf-style formatting */ |
438 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap | |
439 | ||
440 | static inline u64 ceph_ino(struct inode *inode) | |
441 | { | |
442 | return ceph_inode(inode)->i_vino.ino; | |
443 | } | |
444 | static inline u64 ceph_snap(struct inode *inode) | |
445 | { | |
446 | return ceph_inode(inode)->i_vino.snap; | |
447 | } | |
448 | ||
449 | static inline int ceph_ino_compare(struct inode *inode, void *data) | |
450 | { | |
451 | struct ceph_vino *pvino = (struct ceph_vino *)data; | |
452 | struct ceph_inode_info *ci = ceph_inode(inode); | |
453 | return ci->i_vino.ino == pvino->ino && | |
454 | ci->i_vino.snap == pvino->snap; | |
455 | } | |
456 | ||
457 | static inline struct inode *ceph_find_inode(struct super_block *sb, | |
458 | struct ceph_vino vino) | |
459 | { | |
460 | ino_t t = ceph_vino_to_ino(vino); | |
461 | return ilookup5(sb, t, ceph_ino_compare, &vino); | |
462 | } | |
463 | ||
464 | ||
465 | /* | |
466 | * Ceph inode. | |
467 | */ | |
10183a69 YZ |
468 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
469 | #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ | |
470 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ | |
471 | #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ | |
472 | #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ | |
473 | #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ | |
474 | #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ | |
315f2408 | 475 | #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ |
77310320 | 476 | #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ |
13c2b57d | 477 | #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ |
70220ac8 | 478 | #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ |
26544c62 JL |
479 | #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ |
480 | ||
481 | /* | |
482 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode | |
483 | * and then clear it when they start succeeding. Note that we do a lockless | |
484 | * check first, and only take the lock if it looks like it needs to be changed. | |
485 | * The write submission code just takes this as a hint, so we're not too | |
486 | * worried if a few slip through in either direction. | |
487 | */ | |
488 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) | |
489 | { | |
490 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { | |
491 | spin_lock(&ci->i_ceph_lock); | |
492 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; | |
493 | spin_unlock(&ci->i_ceph_lock); | |
494 | } | |
495 | } | |
496 | ||
497 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) | |
498 | { | |
499 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { | |
500 | spin_lock(&ci->i_ceph_lock); | |
501 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; | |
502 | spin_unlock(&ci->i_ceph_lock); | |
503 | } | |
504 | } | |
3d14c5d2 | 505 | |
2f276c51 | 506 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
fdd4e158 YZ |
507 | long long release_count, |
508 | long long ordered_count) | |
de57606c | 509 | { |
fdd4e158 YZ |
510 | smp_mb__before_atomic(); |
511 | atomic64_set(&ci->i_complete_seq[0], release_count); | |
512 | atomic64_set(&ci->i_complete_seq[1], ordered_count); | |
de57606c SW |
513 | } |
514 | ||
2f276c51 | 515 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
de57606c | 516 | { |
fdd4e158 YZ |
517 | atomic64_inc(&ci->i_release_count); |
518 | } | |
519 | ||
520 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) | |
521 | { | |
522 | atomic64_inc(&ci->i_ordered_count); | |
2f276c51 | 523 | } |
de57606c | 524 | |
2f276c51 YZ |
525 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
526 | { | |
fdd4e158 YZ |
527 | return atomic64_read(&ci->i_complete_seq[0]) == |
528 | atomic64_read(&ci->i_release_count); | |
de57606c SW |
529 | } |
530 | ||
70db4f36 YZ |
531 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
532 | { | |
fdd4e158 YZ |
533 | return atomic64_read(&ci->i_complete_seq[0]) == |
534 | atomic64_read(&ci->i_release_count) && | |
535 | atomic64_read(&ci->i_complete_seq[1]) == | |
536 | atomic64_read(&ci->i_ordered_count); | |
70db4f36 YZ |
537 | } |
538 | ||
2f276c51 | 539 | static inline void ceph_dir_clear_complete(struct inode *inode) |
de57606c | 540 | { |
2f276c51 YZ |
541 | __ceph_dir_clear_complete(ceph_inode(inode)); |
542 | } | |
de57606c | 543 | |
70db4f36 | 544 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
2f276c51 | 545 | { |
fdd4e158 | 546 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
de57606c SW |
547 | } |
548 | ||
70db4f36 YZ |
549 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
550 | { | |
fdd4e158 YZ |
551 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
552 | smp_rmb(); | |
70db4f36 YZ |
553 | return ret; |
554 | } | |
de57606c SW |
555 | |
556 | /* find a specific frag @f */ | |
557 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, | |
558 | u32 f); | |
559 | ||
560 | /* | |
561 | * choose fragment for value @v. copy frag content to pfrag, if leaf | |
562 | * exists | |
563 | */ | |
564 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, | |
565 | struct ceph_inode_frag *pfrag, | |
566 | int *found); | |
567 | ||
de57606c SW |
568 | static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) |
569 | { | |
570 | return (struct ceph_dentry_info *)dentry->d_fsdata; | |
571 | } | |
572 | ||
de57606c SW |
573 | /* |
574 | * caps helpers | |
575 | */ | |
576 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) | |
577 | { | |
578 | return !RB_EMPTY_ROOT(&ci->i_caps); | |
579 | } | |
580 | ||
581 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); | |
582 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); | |
583 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, | |
584 | struct ceph_cap *cap); | |
585 | ||
586 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) | |
587 | { | |
588 | int issued; | |
be655596 | 589 | spin_lock(&ci->i_ceph_lock); |
de57606c | 590 | issued = __ceph_caps_issued(ci, NULL); |
be655596 | 591 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
592 | return issued; |
593 | } | |
594 | ||
595 | static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, | |
596 | int touch) | |
597 | { | |
598 | int r; | |
be655596 | 599 | spin_lock(&ci->i_ceph_lock); |
de57606c | 600 | r = __ceph_caps_issued_mask(ci, mask, touch); |
be655596 | 601 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
602 | return r; |
603 | } | |
604 | ||
605 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) | |
606 | { | |
607 | return ci->i_dirty_caps | ci->i_flushing_caps; | |
608 | } | |
f66fd9f0 YZ |
609 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
610 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); | |
611 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, | |
612 | struct ceph_cap_flush **pcf); | |
de57606c | 613 | |
9563f88c YZ |
614 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
615 | struct ceph_cap *ocap, int mask); | |
de57606c SW |
616 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
617 | extern int __ceph_caps_used(struct ceph_inode_info *ci); | |
618 | ||
619 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); | |
620 | ||
621 | /* | |
622 | * wanted, by virtue of open file modes AND cap refs (buffered/cached data) | |
623 | */ | |
624 | static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) | |
625 | { | |
626 | int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); | |
627 | if (w & CEPH_CAP_FILE_BUFFER) | |
628 | w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ | |
629 | return w; | |
630 | } | |
631 | ||
632 | /* what the mds thinks we want */ | |
c1944fed | 633 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
de57606c | 634 | |
37151668 YS |
635 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
636 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); | |
637 | extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); | |
93faca6e | 638 | extern void ceph_reserve_caps(struct ceph_mds_client *mdsc, |
37151668 YS |
639 | struct ceph_cap_reservation *ctx, int need); |
640 | extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, | |
641 | struct ceph_cap_reservation *ctx); | |
3d14c5d2 | 642 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
de57606c | 643 | int *total, int *avail, int *used, |
85ccce43 | 644 | int *reserved, int *min); |
de57606c | 645 | |
de57606c | 646 | |
de57606c SW |
647 | |
648 | /* | |
649 | * we keep buffered readdir results attached to file->private_data | |
650 | */ | |
4918b6d1 | 651 | #define CEPH_F_SYNC 1 |
9cfa1098 | 652 | #define CEPH_F_ATEND 2 |
4918b6d1 | 653 | |
de57606c | 654 | struct ceph_file_info { |
252c6728 SW |
655 | short fmode; /* initialized on open */ |
656 | short flags; /* CEPH_F_* */ | |
de57606c SW |
657 | |
658 | /* readdir: position within the dir */ | |
659 | u32 frag; | |
660 | struct ceph_mds_request *last_readdir; | |
de57606c SW |
661 | |
662 | /* readdir: position within a frag */ | |
f0494206 | 663 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
de57606c | 664 | char *last_name; /* last entry in previous chunk */ |
fdd4e158 YZ |
665 | long long dir_release_count; |
666 | long long dir_ordered_count; | |
667 | int readdir_cache_idx; | |
de57606c SW |
668 | |
669 | /* used for -o dirstat read() on directory thing */ | |
670 | char *dir_info; | |
671 | int dir_info_len; | |
672 | }; | |
673 | ||
fdd4e158 YZ |
674 | struct ceph_readdir_cache_control { |
675 | struct page *page; | |
676 | struct dentry **dentries; | |
677 | int index; | |
678 | }; | |
de57606c | 679 | |
de57606c SW |
680 | /* |
681 | * A "snap realm" describes a subset of the file hierarchy sharing | |
682 | * the same set of snapshots that apply to it. The realms themselves | |
683 | * are organized into a hierarchy, such that children inherit (some of) | |
684 | * the snapshots of their parents. | |
685 | * | |
686 | * All inodes within the realm that have capabilities are linked into a | |
687 | * per-realm list. | |
688 | */ | |
689 | struct ceph_snap_realm { | |
690 | u64 ino; | |
691 | atomic_t nref; | |
a105f00c SW |
692 | struct rb_node node; |
693 | ||
de57606c SW |
694 | u64 created, seq; |
695 | u64 parent_ino; | |
696 | u64 parent_since; /* snapid when our current parent became so */ | |
697 | ||
698 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ | |
aa711ee3 | 699 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
de57606c | 700 | u64 *snaps; /* snaps specific to this realm */ |
aa711ee3 | 701 | u32 num_snaps; |
de57606c SW |
702 | |
703 | struct ceph_snap_realm *parent; | |
704 | struct list_head children; /* list of child realms */ | |
705 | struct list_head child_item; | |
706 | ||
707 | struct list_head empty_item; /* if i have ref==0 */ | |
708 | ||
ae00d4f3 SW |
709 | struct list_head dirty_item; /* if realm needs new context */ |
710 | ||
de57606c SW |
711 | /* the current set of snaps for this realm */ |
712 | struct ceph_snap_context *cached_context; | |
713 | ||
714 | struct list_head inodes_with_caps; | |
715 | spinlock_t inodes_with_caps_lock; | |
716 | }; | |
717 | ||
3d14c5d2 | 718 | static inline int default_congestion_kb(void) |
de57606c | 719 | { |
3d14c5d2 YS |
720 | int congestion_kb; |
721 | ||
722 | /* | |
723 | * Copied from NFS | |
724 | * | |
725 | * congestion size, scale with available memory. | |
726 | * | |
727 | * 64MB: 8192k | |
728 | * 128MB: 11585k | |
729 | * 256MB: 16384k | |
730 | * 512MB: 23170k | |
731 | * 1GB: 32768k | |
732 | * 2GB: 46340k | |
733 | * 4GB: 65536k | |
734 | * 8GB: 92681k | |
735 | * 16GB: 131072k | |
736 | * | |
737 | * This allows larger machines to have larger/more transfers. | |
738 | * Limit the default to 256M | |
739 | */ | |
740 | congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); | |
741 | if (congestion_kb > 256*1024) | |
742 | congestion_kb = 256*1024; | |
743 | ||
744 | return congestion_kb; | |
de57606c SW |
745 | } |
746 | ||
747 | ||
748 | ||
749 | /* snap.c */ | |
750 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, | |
751 | u64 ino); | |
752 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, | |
753 | struct ceph_snap_realm *realm); | |
754 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, | |
755 | struct ceph_snap_realm *realm); | |
756 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, | |
982d6011 YZ |
757 | void *p, void *e, bool deletion, |
758 | struct ceph_snap_realm **realm_ret); | |
de57606c | 759 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
2600d2dd | 760 | struct ceph_mds_session *session, |
de57606c | 761 | struct ceph_msg *msg); |
fc837c8f | 762 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
de57606c SW |
763 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
764 | struct ceph_cap_snap *capsnap); | |
765 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); | |
766 | ||
767 | /* | |
768 | * a cap_snap is "pending" if it is still awaiting an in-progress | |
769 | * sync write (that may/may not still update size, mtime, etc.). | |
770 | */ | |
771 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) | |
772 | { | |
773 | return !list_empty(&ci->i_cap_snaps) && | |
86056090 YZ |
774 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
775 | ci_item)->writing; | |
de57606c SW |
776 | } |
777 | ||
de57606c SW |
778 | /* inode.c */ |
779 | extern const struct inode_operations ceph_file_iops; | |
780 | ||
781 | extern struct inode *ceph_alloc_inode(struct super_block *sb); | |
782 | extern void ceph_destroy_inode(struct inode *inode); | |
9f12bd11 | 783 | extern int ceph_drop_inode(struct inode *inode); |
de57606c SW |
784 | |
785 | extern struct inode *ceph_get_inode(struct super_block *sb, | |
786 | struct ceph_vino vino); | |
787 | extern struct inode *ceph_get_snapdir(struct inode *parent); | |
788 | extern int ceph_fill_file_size(struct inode *inode, int issued, | |
789 | u32 truncate_seq, u64 truncate_size, u64 size); | |
790 | extern void ceph_fill_file_time(struct inode *inode, int issued, | |
791 | u64 time_warp_seq, struct timespec *ctime, | |
792 | struct timespec *mtime, struct timespec *atime); | |
793 | extern int ceph_fill_trace(struct super_block *sb, | |
f5a03b08 | 794 | struct ceph_mds_request *req); |
de57606c SW |
795 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
796 | struct ceph_mds_session *session); | |
797 | ||
798 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); | |
799 | ||
efb0ca76 | 800 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
b415bf4f | 801 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
3c6f6b79 SW |
802 | extern void ceph_queue_vmtruncate(struct inode *inode); |
803 | ||
804 | extern void ceph_queue_invalidate(struct inode *inode); | |
805 | extern void ceph_queue_writeback(struct inode *inode); | |
de57606c | 806 | |
01deead0 YZ |
807 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
808 | int mask, bool force); | |
809 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) | |
810 | { | |
811 | return __ceph_do_getattr(inode, NULL, mask, force); | |
812 | } | |
10556cb2 | 813 | extern int ceph_permission(struct inode *inode, int mask); |
a26fecca | 814 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
de57606c | 815 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
a528d35e DH |
816 | extern int ceph_getattr(const struct path *path, struct kstat *stat, |
817 | u32 request_mask, unsigned int flags); | |
de57606c SW |
818 | |
819 | /* xattr.c */ | |
a26fecca | 820 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
7221fe4c | 821 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
de57606c | 822 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
de57606c SW |
823 | extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
824 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); | |
3ce6cd12 AE |
825 | extern void __init ceph_xattr_init(void); |
826 | extern void ceph_xattr_exit(void); | |
b1ee94aa | 827 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
de57606c | 828 | |
315f2408 YZ |
829 | #ifdef CONFIG_SECURITY |
830 | extern bool ceph_security_xattr_deadlock(struct inode *in); | |
831 | extern bool ceph_security_xattr_wanted(struct inode *in); | |
832 | #else | |
833 | static inline bool ceph_security_xattr_deadlock(struct inode *in) | |
834 | { | |
835 | return false; | |
836 | } | |
837 | static inline bool ceph_security_xattr_wanted(struct inode *in) | |
838 | { | |
839 | return false; | |
840 | } | |
841 | #endif | |
842 | ||
7221fe4c | 843 | /* acl.c */ |
b1ee94aa YZ |
844 | struct ceph_acls_info { |
845 | void *default_acl; | |
846 | void *acl; | |
847 | struct ceph_pagelist *pagelist; | |
848 | }; | |
7221fe4c GZ |
849 | |
850 | #ifdef CONFIG_CEPH_FS_POSIX_ACL | |
851 | ||
852 | struct posix_acl *ceph_get_acl(struct inode *, int); | |
72466d0b | 853 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
b1ee94aa YZ |
854 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
855 | struct ceph_acls_info *info); | |
856 | void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); | |
857 | void ceph_release_acls_info(struct ceph_acls_info *info); | |
c969d9bf GZ |
858 | |
859 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
860 | { | |
861 | forget_all_cached_acls(inode); | |
862 | } | |
7221fe4c GZ |
863 | |
864 | #else | |
865 | ||
866 | #define ceph_get_acl NULL | |
72466d0b | 867 | #define ceph_set_acl NULL |
7221fe4c | 868 | |
b1ee94aa YZ |
869 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
870 | struct ceph_acls_info *info) | |
7221fe4c GZ |
871 | { |
872 | return 0; | |
873 | } | |
b1ee94aa YZ |
874 | static inline void ceph_init_inode_acls(struct inode *inode, |
875 | struct ceph_acls_info *info) | |
876 | { | |
877 | } | |
878 | static inline void ceph_release_acls_info(struct ceph_acls_info *info) | |
879 | { | |
880 | } | |
7221fe4c GZ |
881 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
882 | { | |
883 | return 0; | |
884 | } | |
885 | ||
886 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
887 | { | |
888 | } | |
889 | ||
890 | #endif | |
891 | ||
de57606c SW |
892 | /* caps.c */ |
893 | extern const char *ceph_cap_string(int c); | |
894 | extern void ceph_handle_caps(struct ceph_mds_session *session, | |
895 | struct ceph_msg *msg); | |
d9df2783 YZ |
896 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
897 | struct ceph_cap_reservation *ctx); | |
898 | extern void ceph_add_cap(struct inode *inode, | |
899 | struct ceph_mds_session *session, u64 cap_id, | |
900 | int fmode, unsigned issued, unsigned wanted, | |
901 | unsigned cap, unsigned seq, u64 realmino, int flags, | |
902 | struct ceph_cap **new_cap); | |
a096b09a | 903 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
37151668 YS |
904 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
905 | struct ceph_cap *cap); | |
9215aeea | 906 | extern int ceph_is_any_caps(struct inode *inode); |
de57606c SW |
907 | |
908 | extern void ceph_queue_caps_release(struct inode *inode); | |
f1a3d572 | 909 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
02c24a82 JB |
910 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
911 | int datasync); | |
e548e9b9 YZ |
912 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
913 | struct ceph_mds_session *session); | |
de57606c SW |
914 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
915 | struct ceph_mds_session *session); | |
2bc50259 GF |
916 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
917 | int mds); | |
de57606c SW |
918 | extern int ceph_get_cap_mds(struct inode *inode); |
919 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); | |
920 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); | |
921 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, | |
922 | struct ceph_snap_context *snapc); | |
ed9b430c YZ |
923 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
924 | struct ceph_mds_session **psession); | |
efb0ca76 | 925 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
de57606c SW |
926 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
927 | struct ceph_mds_session *session); | |
afcdaea3 SW |
928 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
929 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); | |
de57606c SW |
930 | |
931 | extern int ceph_encode_inode_release(void **p, struct inode *inode, | |
932 | int mds, int drop, int unless, int force); | |
933 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, | |
ca6c8ae0 | 934 | struct inode *dir, |
de57606c SW |
935 | int mds, int drop, int unless); |
936 | ||
937 | extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, | |
31c542a1 | 938 | loff_t endoff, int *got, struct page **pinned_page); |
2b1ac852 YZ |
939 | extern int ceph_try_get_caps(struct ceph_inode_info *ci, |
940 | int need, int want, int *got); | |
de57606c SW |
941 | |
942 | /* for counting open files by mode */ | |
774a6a11 | 943 | extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); |
de57606c SW |
944 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); |
945 | ||
946 | /* addr.c */ | |
947 | extern const struct address_space_operations ceph_aops; | |
948 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); | |
10183a69 YZ |
949 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
950 | extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); | |
951 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); | |
de57606c SW |
952 | |
953 | /* file.c */ | |
954 | extern const struct file_operations ceph_file_fops; | |
9e0eb85d | 955 | |
77310320 | 956 | extern int ceph_renew_caps(struct inode *inode); |
de57606c | 957 | extern int ceph_open(struct inode *inode, struct file *file); |
5ef50c3b SW |
958 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
959 | struct file *file, unsigned flags, umode_t mode, | |
960 | int *opened); | |
de57606c | 961 | extern int ceph_release(struct inode *inode, struct file *filp); |
31c542a1 YZ |
962 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
963 | char *data, size_t len); | |
55f2a045 | 964 | |
de57606c SW |
965 | /* dir.c */ |
966 | extern const struct file_operations ceph_dir_fops; | |
38c48b5f | 967 | extern const struct file_operations ceph_snapdir_fops; |
de57606c | 968 | extern const struct inode_operations ceph_dir_iops; |
38c48b5f | 969 | extern const struct inode_operations ceph_snapdir_iops; |
18fc8abd | 970 | extern const struct dentry_operations ceph_dentry_ops; |
de57606c | 971 | |
f3c4ebe6 | 972 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
de57606c | 973 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
468640e3 SW |
974 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
975 | struct dentry *dentry, int err); | |
de57606c SW |
976 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
977 | struct dentry *dentry, int err); | |
978 | ||
979 | extern void ceph_dentry_lru_add(struct dentry *dn); | |
980 | extern void ceph_dentry_lru_touch(struct dentry *dn); | |
981 | extern void ceph_dentry_lru_del(struct dentry *dn); | |
81a6cf2d | 982 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
e5f86dc3 | 983 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
fdd4e158 | 984 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
de57606c | 985 | |
de57606c SW |
986 | /* ioctl.c */ |
987 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); | |
988 | ||
989 | /* export.c */ | |
990 | extern const struct export_operations ceph_export_ops; | |
991 | ||
40819f6f | 992 | /* locks.c */ |
eb13e832 | 993 | extern __init void ceph_flock_init(void); |
40819f6f GF |
994 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
995 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); | |
996 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); | |
39be95e9 JS |
997 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
998 | struct ceph_filelock *flocks, | |
999 | int num_fcntl_locks, | |
1000 | int num_flock_locks); | |
1001 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, | |
1002 | struct ceph_pagelist *pagelist, | |
1003 | int num_fcntl_locks, int num_flock_locks); | |
40819f6f GF |
1004 | extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c); |
1005 | ||
3d14c5d2 YS |
1006 | /* debugfs.c */ |
1007 | extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); | |
1008 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); | |
1009 | ||
de57606c | 1010 | #endif /* _FS_CEPH_SUPER_H */ |