]>
Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
de57606c SW |
2 | #ifndef _FS_CEPH_SUPER_H |
3 | #define _FS_CEPH_SUPER_H | |
4 | ||
3d14c5d2 | 5 | #include <linux/ceph/ceph_debug.h> |
de57606c SW |
6 | |
7 | #include <asm/unaligned.h> | |
8 | #include <linux/backing-dev.h> | |
9 | #include <linux/completion.h> | |
10 | #include <linux/exportfs.h> | |
11 | #include <linux/fs.h> | |
12 | #include <linux/mempool.h> | |
13 | #include <linux/pagemap.h> | |
14 | #include <linux/wait.h> | |
f1a3d572 | 15 | #include <linux/writeback.h> |
5a0e3ad6 | 16 | #include <linux/slab.h> |
c969d9bf | 17 | #include <linux/posix_acl.h> |
805692d0 | 18 | #include <linux/refcount.h> |
de57606c | 19 | |
3d14c5d2 | 20 | #include <linux/ceph/libceph.h> |
de57606c | 21 | |
99ccbd22 MT |
22 | #ifdef CONFIG_CEPH_FSCACHE |
23 | #include <linux/fscache.h> | |
24 | #endif | |
25 | ||
de57606c SW |
26 | /* f_type in struct statfs */ |
27 | #define CEPH_SUPER_MAGIC 0x00c36400 | |
28 | ||
29 | /* large granularity for statfs utilization stats to facilitate | |
30 | * large volume sizes on 32-bit machines. */ | |
92a49fb0 | 31 | #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ |
de57606c SW |
32 | #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) |
33 | ||
3d14c5d2 YS |
34 | #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ |
35 | #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ | |
36 | #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ | |
ad1fee96 | 37 | #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ |
a40dc6cc | 38 | #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ |
99ccbd22 | 39 | #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ |
10183a69 | 40 | #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ |
e9e427f0 | 41 | #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ |
6a259382 | 42 | |
133e9156 | 43 | #define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE |
de57606c | 44 | |
3d14c5d2 YS |
45 | #define ceph_set_mount_opt(fsc, opt) \ |
46 | (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; | |
47 | #define ceph_test_mount_opt(fsc, opt) \ | |
48 | (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) | |
de57606c | 49 | |
aa187926 YZ |
50 | /* max size of osd read request, limited by libceph */ |
51 | #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN | |
95cca2b4 YZ |
52 | /* osd has a configurable limitaion of max write size. |
53 | * CEPH_MSG_MAX_DATA_LEN should be small enough. */ | |
54 | #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN | |
7c94ba27 | 55 | #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ |
3d14c5d2 YS |
56 | #define CEPH_MAX_READDIR_DEFAULT 1024 |
57 | #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) | |
58 | #define CEPH_SNAPDIRNAME_DEFAULT ".snap" | |
de57606c | 59 | |
4214fb15 YZ |
60 | /* |
61 | * Delay telling the MDS we no longer want caps, in case we reopen | |
62 | * the file. Delay a minimum amount of time, even if we send a cap | |
63 | * message for some other reason. Otherwise, take the oppotunity to | |
64 | * update the mds to avoid sending another message later. | |
65 | */ | |
66 | #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ | |
67 | #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ | |
68 | ||
3d14c5d2 | 69 | struct ceph_mount_options { |
6e19a16e | 70 | int flags; |
3d14c5d2 YS |
71 | int sb_flags; |
72 | ||
83817e35 SW |
73 | int wsize; /* max write size */ |
74 | int rsize; /* max read size */ | |
75 | int rasize; /* max readahead */ | |
6e19a16e SW |
76 | int congestion_kb; /* max writeback in flight */ |
77 | int caps_wanted_delay_min, caps_wanted_delay_max; | |
23804d91 SW |
78 | int max_readdir; /* max readdir result (entires) */ |
79 | int max_readdir_bytes; /* max readdir result (bytes) */ | |
de57606c | 80 | |
3d14c5d2 YS |
81 | /* |
82 | * everything above this point can be memcmp'd; everything below | |
83 | * is handled in compare_mount_options() | |
84 | */ | |
de57606c | 85 | |
3d14c5d2 | 86 | char *snapdir_name; /* default ".snap" */ |
430afbad | 87 | char *mds_namespace; /* default NULL */ |
3f384954 | 88 | char *server_path; /* default "/" */ |
1d8f8360 | 89 | char *fscache_uniq; /* default NULL */ |
3d14c5d2 | 90 | }; |
de57606c | 91 | |
3d14c5d2 | 92 | struct ceph_fs_client { |
de57606c SW |
93 | struct super_block *sb; |
94 | ||
3d14c5d2 YS |
95 | struct ceph_mount_options *mount_options; |
96 | struct ceph_client *client; | |
de57606c | 97 | |
3d14c5d2 | 98 | unsigned long mount_state; |
85ccce43 SW |
99 | int min_caps; /* min caps i added */ |
100 | ||
3d14c5d2 | 101 | struct ceph_mds_client *mdsc; |
de57606c SW |
102 | |
103 | /* writeback */ | |
104 | mempool_t *wb_pagevec_pool; | |
1de82206 YZ |
105 | atomic_long_t writeback_count; |
106 | ||
d91016e8 | 107 | struct workqueue_struct *inode_wq; |
1de82206 | 108 | struct workqueue_struct *cap_wq; |
de57606c | 109 | |
0743304d | 110 | #ifdef CONFIG_DEBUG_FS |
3d14c5d2 | 111 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
2baba250 | 112 | struct dentry *debugfs_congestion_kb; |
06edf046 | 113 | struct dentry *debugfs_bdi; |
3d14c5d2 | 114 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
14ed9703 | 115 | struct dentry *debugfs_mds_sessions; |
0743304d | 116 | #endif |
99ccbd22 MT |
117 | |
118 | #ifdef CONFIG_CEPH_FSCACHE | |
119 | struct fscache_cookie *fscache; | |
99ccbd22 | 120 | #endif |
de57606c SW |
121 | }; |
122 | ||
3d14c5d2 | 123 | |
de57606c SW |
124 | /* |
125 | * File i/o capability. This tracks shared state with the metadata | |
126 | * server that allows us to cache or writeback attributes or to read | |
127 | * and write data. For any given inode, we should have one or more | |
128 | * capabilities, one issued by each metadata server, and our | |
129 | * cumulative access is the OR of all issued capabilities. | |
130 | * | |
131 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds | |
132 | * session capability lists. | |
133 | */ | |
134 | struct ceph_cap { | |
135 | struct ceph_inode_info *ci; | |
136 | struct rb_node ci_node; /* per-ci cap tree */ | |
137 | struct ceph_mds_session *session; | |
138 | struct list_head session_caps; /* per-session caplist */ | |
de57606c | 139 | u64 cap_id; /* unique cap id (mds provided) */ |
745a8e3b YZ |
140 | union { |
141 | /* in-use caps */ | |
142 | struct { | |
143 | int issued; /* latest, from the mds */ | |
144 | int implemented; /* implemented superset of | |
145 | issued (for revocation) */ | |
146 | int mds, mds_wanted; | |
147 | }; | |
148 | /* caps to release */ | |
149 | struct { | |
150 | u64 cap_ino; | |
151 | int queue_release; | |
152 | }; | |
153 | }; | |
685f9a5d SW |
154 | u32 seq, issue_seq, mseq; |
155 | u32 cap_gen; /* active/stale cycle */ | |
de57606c SW |
156 | unsigned long last_used; |
157 | struct list_head caps_item; | |
158 | }; | |
159 | ||
160 | #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ | |
161 | #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ | |
162 | #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ | |
163 | ||
0e294387 YZ |
164 | struct ceph_cap_flush { |
165 | u64 tid; | |
166 | int caps; /* 0 means capsnap */ | |
c8799fc4 | 167 | bool wake; /* wake up flush waiters when finish ? */ |
0e294387 YZ |
168 | struct list_head g_list; // global |
169 | struct list_head i_list; // per inode | |
170 | }; | |
171 | ||
de57606c SW |
172 | /* |
173 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, | |
174 | * we first complete any in-process sync writes and writeback any dirty | |
175 | * data before flushing the snapped state (tracked here) back to the MDS. | |
176 | */ | |
177 | struct ceph_cap_snap { | |
805692d0 | 178 | refcount_t nref; |
0e294387 | 179 | struct list_head ci_item; |
de57606c | 180 | |
0e294387 YZ |
181 | struct ceph_cap_flush cap_flush; |
182 | ||
183 | u64 follows; | |
de57606c SW |
184 | int issued, dirty; |
185 | struct ceph_snap_context *context; | |
186 | ||
5706b27d | 187 | umode_t mode; |
05cb11c1 EB |
188 | kuid_t uid; |
189 | kgid_t gid; | |
de57606c | 190 | |
4a625be4 | 191 | struct ceph_buffer *xattr_blob; |
de57606c SW |
192 | u64 xattr_version; |
193 | ||
194 | u64 size; | |
195 | struct timespec mtime, atime, ctime; | |
196 | u64 time_warp_seq; | |
5f743e45 YZ |
197 | u64 truncate_size; |
198 | u32 truncate_seq; | |
de57606c SW |
199 | int writing; /* a sync write is still in progress */ |
200 | int dirty_pages; /* dirty pages awaiting writeback */ | |
e20d258d | 201 | bool inline_data; |
86056090 | 202 | bool need_flush; |
de57606c SW |
203 | }; |
204 | ||
205 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) | |
206 | { | |
805692d0 | 207 | if (refcount_dec_and_test(&capsnap->nref)) { |
4a625be4 SW |
208 | if (capsnap->xattr_blob) |
209 | ceph_buffer_put(capsnap->xattr_blob); | |
de57606c | 210 | kfree(capsnap); |
4a625be4 | 211 | } |
de57606c SW |
212 | } |
213 | ||
214 | /* | |
215 | * The frag tree describes how a directory is fragmented, potentially across | |
216 | * multiple metadata servers. It is also used to indicate points where | |
217 | * metadata authority is delegated, and whether/where metadata is replicated. | |
218 | * | |
219 | * A _leaf_ frag will be present in the i_fragtree IFF there is | |
220 | * delegation info. That is, if mds >= 0 || ndist > 0. | |
221 | */ | |
222 | #define CEPH_MAX_DIRFRAG_REP 4 | |
223 | ||
224 | struct ceph_inode_frag { | |
225 | struct rb_node node; | |
226 | ||
227 | /* fragtree state */ | |
228 | u32 frag; | |
229 | int split_by; /* i.e. 2^(split_by) children */ | |
230 | ||
231 | /* delegation and replication info */ | |
232 | int mds; /* -1 if same authority as parent */ | |
233 | int ndist; /* >0 if replicated */ | |
234 | int dist[CEPH_MAX_DIRFRAG_REP]; | |
235 | }; | |
236 | ||
237 | /* | |
238 | * We cache inode xattrs as an encoded blob until they are first used, | |
239 | * at which point we parse them into an rbtree. | |
240 | */ | |
241 | struct ceph_inode_xattr { | |
242 | struct rb_node node; | |
243 | ||
244 | const char *name; | |
245 | int name_len; | |
246 | const char *val; | |
247 | int val_len; | |
248 | int dirty; | |
249 | ||
250 | int should_free_name; | |
251 | int should_free_val; | |
252 | }; | |
253 | ||
3d14c5d2 YS |
254 | /* |
255 | * Ceph dentry state | |
256 | */ | |
257 | struct ceph_dentry_info { | |
258 | struct ceph_mds_session *lease_session; | |
7389f7b2 YZ |
259 | int lease_shared_gen; |
260 | u32 lease_gen; | |
3d14c5d2 YS |
261 | u32 lease_seq; |
262 | unsigned long lease_renew_after, lease_renew_from; | |
263 | struct list_head lru; | |
264 | struct dentry *dentry; | |
9b16f03c | 265 | unsigned long time; |
3d14c5d2 YS |
266 | u64 offset; |
267 | }; | |
268 | ||
de57606c SW |
269 | struct ceph_inode_xattrs_info { |
270 | /* | |
271 | * (still encoded) xattr blob. we avoid the overhead of parsing | |
272 | * this until someone actually calls getxattr, etc. | |
273 | * | |
274 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == | |
275 | * NULL means we don't know. | |
276 | */ | |
277 | struct ceph_buffer *blob, *prealloc_blob; | |
278 | ||
279 | struct rb_root index; | |
280 | bool dirty; | |
281 | int count; | |
282 | int names_size; | |
283 | int vals_size; | |
284 | u64 version, index_version; | |
285 | }; | |
286 | ||
287 | /* | |
288 | * Ceph inode. | |
289 | */ | |
de57606c SW |
290 | struct ceph_inode_info { |
291 | struct ceph_vino i_vino; /* ceph ino + snap */ | |
292 | ||
be655596 SW |
293 | spinlock_t i_ceph_lock; |
294 | ||
de57606c | 295 | u64 i_version; |
31c542a1 | 296 | u64 i_inline_version; |
de57606c SW |
297 | u32 i_time_warp_seq; |
298 | ||
299 | unsigned i_ceph_flags; | |
fdd4e158 YZ |
300 | atomic64_t i_release_count; |
301 | atomic64_t i_ordered_count; | |
302 | atomic64_t i_complete_seq[2]; | |
de57606c | 303 | |
6c0f3af7 | 304 | struct ceph_dir_layout i_dir_layout; |
de57606c SW |
305 | struct ceph_file_layout i_layout; |
306 | char *i_symlink; | |
307 | ||
308 | /* for dirs */ | |
309 | struct timespec i_rctime; | |
310 | u64 i_rbytes, i_rfiles, i_rsubdirs; | |
311 | u64 i_files, i_subdirs; | |
de57606c | 312 | |
7370e8a4 LH |
313 | /* quotas */ |
314 | u64 i_max_bytes, i_max_files; | |
315 | ||
de57606c | 316 | struct rb_root i_fragtree; |
1b1bc16d | 317 | int i_fragtree_nsplits; |
de57606c SW |
318 | struct mutex i_fragtree_mutex; |
319 | ||
320 | struct ceph_inode_xattrs_info i_xattrs; | |
321 | ||
be655596 | 322 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
de57606c SW |
323 | * s_mutex. */ |
324 | struct rb_root i_caps; /* cap list */ | |
325 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ | |
326 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ | |
327 | struct list_head i_dirty_item, i_flushing_item; | |
de57606c SW |
328 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
329 | * overlapping, pipelined cap flushes to the mds. we can probably | |
330 | * reduce the tid to 8 bits if we're concerned about inode size. */ | |
f66fd9f0 | 331 | struct ceph_cap_flush *i_prealloc_cap_flush; |
e4500b5e | 332 | struct list_head i_cap_flush_list; |
de57606c SW |
333 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
334 | unsigned long i_hold_caps_min; /* jiffies */ | |
335 | unsigned long i_hold_caps_max; /* jiffies */ | |
336 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ | |
de57606c SW |
337 | struct ceph_cap_reservation i_cap_migration_resv; |
338 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ | |
7d8cb26d SW |
339 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
340 | dirty|flushing caps */ | |
de57606c SW |
341 | unsigned i_snap_caps; /* cap bits for snapped files */ |
342 | ||
774a6a11 | 343 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
de57606c | 344 | |
b0d7c223 | 345 | struct mutex i_truncate_mutex; |
de57606c SW |
346 | u32 i_truncate_seq; /* last truncate to smaller size */ |
347 | u64 i_truncate_size; /* and the size we last truncated down to */ | |
348 | int i_truncate_pending; /* still need to call vmtruncate */ | |
349 | ||
350 | u64 i_max_size; /* max file size authorized by mds */ | |
351 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ | |
352 | u64 i_wanted_max_size; /* offset we'd like to write too */ | |
353 | u64 i_requested_max_size; /* max_size we've requested */ | |
354 | ||
355 | /* held references to caps */ | |
356 | int i_pin_ref; | |
d3d0720d | 357 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; |
de57606c | 358 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
89aa5930 | 359 | atomic_t i_filelock_ref; |
7389f7b2 | 360 | atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ |
cd045cb4 | 361 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
de57606c SW |
362 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
363 | ||
de57606c | 364 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
68cd5b4b | 365 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
de57606c SW |
366 | spinlock_t i_unsafe_lock; |
367 | ||
368 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ | |
369 | int i_snap_realm_counter; /* snap realm (if caps) */ | |
370 | struct list_head i_snap_realm_item; | |
371 | struct list_head i_snap_flush_item; | |
372 | ||
d91016e8 YZ |
373 | struct work_struct i_work; |
374 | unsigned long i_work_mask; | |
de57606c | 375 | |
99ccbd22 MT |
376 | #ifdef CONFIG_CEPH_FSCACHE |
377 | struct fscache_cookie *fscache; | |
f7f7e7a0 | 378 | u32 i_fscache_gen; |
99ccbd22 | 379 | #endif |
de57606c SW |
380 | struct inode vfs_inode; /* at end */ |
381 | }; | |
382 | ||
383 | static inline struct ceph_inode_info *ceph_inode(struct inode *inode) | |
384 | { | |
fbbccec9 | 385 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
de57606c SW |
386 | } |
387 | ||
ad1fee96 YS |
388 | static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) |
389 | { | |
390 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; | |
391 | } | |
392 | ||
393 | static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) | |
394 | { | |
395 | return (struct ceph_fs_client *)sb->s_fs_info; | |
396 | } | |
397 | ||
3d14c5d2 YS |
398 | static inline struct ceph_vino ceph_vino(struct inode *inode) |
399 | { | |
400 | return ceph_inode(inode)->i_vino; | |
401 | } | |
402 | ||
403 | /* | |
404 | * ino_t is <64 bits on many architectures, blech. | |
405 | * | |
ad1fee96 YS |
406 | * i_ino (kernel inode) st_ino (userspace) |
407 | * i386 32 32 | |
408 | * x86_64+ino32 64 32 | |
409 | * x86_64 64 64 | |
410 | */ | |
3310f754 | 411 | static inline u32 ceph_ino_to_ino32(__u64 vino) |
ad1fee96 | 412 | { |
3310f754 AO |
413 | u32 ino = vino & 0xffffffff; |
414 | ino ^= vino >> 32; | |
ad1fee96 | 415 | if (!ino) |
a661fc56 | 416 | ino = 2; |
ad1fee96 YS |
417 | return ino; |
418 | } | |
419 | ||
420 | /* | |
421 | * kernel i_ino value | |
3d14c5d2 YS |
422 | */ |
423 | static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) | |
424 | { | |
3d14c5d2 | 425 | #if BITS_PER_LONG == 32 |
3310f754 AO |
426 | return ceph_ino_to_ino32(vino.ino); |
427 | #else | |
428 | return (ino_t)vino.ino; | |
3d14c5d2 | 429 | #endif |
3d14c5d2 YS |
430 | } |
431 | ||
ad1fee96 YS |
432 | /* |
433 | * user-visible ino (stat, filldir) | |
434 | */ | |
435 | #if BITS_PER_LONG == 32 | |
436 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
437 | { | |
438 | return ino; | |
439 | } | |
440 | #else | |
441 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
442 | { | |
443 | if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) | |
444 | ino = ceph_ino_to_ino32(ino); | |
445 | return ino; | |
446 | } | |
447 | #endif | |
448 | ||
449 | ||
3d14c5d2 YS |
450 | /* for printf-style formatting */ |
451 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap | |
452 | ||
453 | static inline u64 ceph_ino(struct inode *inode) | |
454 | { | |
455 | return ceph_inode(inode)->i_vino.ino; | |
456 | } | |
457 | static inline u64 ceph_snap(struct inode *inode) | |
458 | { | |
459 | return ceph_inode(inode)->i_vino.snap; | |
460 | } | |
461 | ||
462 | static inline int ceph_ino_compare(struct inode *inode, void *data) | |
463 | { | |
464 | struct ceph_vino *pvino = (struct ceph_vino *)data; | |
465 | struct ceph_inode_info *ci = ceph_inode(inode); | |
466 | return ci->i_vino.ino == pvino->ino && | |
467 | ci->i_vino.snap == pvino->snap; | |
468 | } | |
469 | ||
470 | static inline struct inode *ceph_find_inode(struct super_block *sb, | |
471 | struct ceph_vino vino) | |
472 | { | |
473 | ino_t t = ceph_vino_to_ino(vino); | |
474 | return ilookup5(sb, t, ceph_ino_compare, &vino); | |
475 | } | |
476 | ||
477 | ||
478 | /* | |
479 | * Ceph inode. | |
480 | */ | |
10183a69 YZ |
481 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
482 | #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ | |
483 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ | |
484 | #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ | |
485 | #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ | |
486 | #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ | |
487 | #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ | |
315f2408 | 488 | #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ |
77310320 | 489 | #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ |
13c2b57d | 490 | #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ |
70220ac8 | 491 | #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ |
26544c62 | 492 | #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ |
b3f8d68f YZ |
493 | #define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */ |
494 | ||
26544c62 | 495 | |
d91016e8 YZ |
496 | /* |
497 | * Masks of ceph inode work. | |
498 | */ | |
499 | #define CEPH_I_WORK_WRITEBACK 0 /* writeback */ | |
500 | #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */ | |
501 | #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */ | |
502 | ||
26544c62 JL |
503 | /* |
504 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode | |
505 | * and then clear it when they start succeeding. Note that we do a lockless | |
506 | * check first, and only take the lock if it looks like it needs to be changed. | |
507 | * The write submission code just takes this as a hint, so we're not too | |
508 | * worried if a few slip through in either direction. | |
509 | */ | |
510 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) | |
511 | { | |
512 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { | |
513 | spin_lock(&ci->i_ceph_lock); | |
514 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; | |
515 | spin_unlock(&ci->i_ceph_lock); | |
516 | } | |
517 | } | |
518 | ||
519 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) | |
520 | { | |
521 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { | |
522 | spin_lock(&ci->i_ceph_lock); | |
523 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; | |
524 | spin_unlock(&ci->i_ceph_lock); | |
525 | } | |
526 | } | |
3d14c5d2 | 527 | |
2f276c51 | 528 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
fdd4e158 YZ |
529 | long long release_count, |
530 | long long ordered_count) | |
de57606c | 531 | { |
6a1ccdbd AP |
532 | /* |
533 | * Makes sure operations that setup readdir cache (update page | |
534 | * cache and i_size) are strongly ordered w.r.t. the following | |
535 | * atomic64_set() operations. | |
536 | */ | |
537 | smp_mb(); | |
fdd4e158 YZ |
538 | atomic64_set(&ci->i_complete_seq[0], release_count); |
539 | atomic64_set(&ci->i_complete_seq[1], ordered_count); | |
de57606c SW |
540 | } |
541 | ||
2f276c51 | 542 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
de57606c | 543 | { |
fdd4e158 YZ |
544 | atomic64_inc(&ci->i_release_count); |
545 | } | |
546 | ||
547 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) | |
548 | { | |
549 | atomic64_inc(&ci->i_ordered_count); | |
2f276c51 | 550 | } |
de57606c | 551 | |
2f276c51 YZ |
552 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
553 | { | |
fdd4e158 YZ |
554 | return atomic64_read(&ci->i_complete_seq[0]) == |
555 | atomic64_read(&ci->i_release_count); | |
de57606c SW |
556 | } |
557 | ||
70db4f36 YZ |
558 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
559 | { | |
fdd4e158 YZ |
560 | return atomic64_read(&ci->i_complete_seq[0]) == |
561 | atomic64_read(&ci->i_release_count) && | |
562 | atomic64_read(&ci->i_complete_seq[1]) == | |
563 | atomic64_read(&ci->i_ordered_count); | |
70db4f36 YZ |
564 | } |
565 | ||
2f276c51 | 566 | static inline void ceph_dir_clear_complete(struct inode *inode) |
de57606c | 567 | { |
2f276c51 YZ |
568 | __ceph_dir_clear_complete(ceph_inode(inode)); |
569 | } | |
de57606c | 570 | |
70db4f36 | 571 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
2f276c51 | 572 | { |
fdd4e158 | 573 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
de57606c SW |
574 | } |
575 | ||
70db4f36 YZ |
576 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
577 | { | |
fdd4e158 YZ |
578 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
579 | smp_rmb(); | |
70db4f36 YZ |
580 | return ret; |
581 | } | |
de57606c SW |
582 | |
583 | /* find a specific frag @f */ | |
584 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, | |
585 | u32 f); | |
586 | ||
587 | /* | |
588 | * choose fragment for value @v. copy frag content to pfrag, if leaf | |
589 | * exists | |
590 | */ | |
591 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, | |
592 | struct ceph_inode_frag *pfrag, | |
593 | int *found); | |
594 | ||
de57606c SW |
595 | static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) |
596 | { | |
597 | return (struct ceph_dentry_info *)dentry->d_fsdata; | |
598 | } | |
599 | ||
de57606c SW |
600 | /* |
601 | * caps helpers | |
602 | */ | |
603 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) | |
604 | { | |
605 | return !RB_EMPTY_ROOT(&ci->i_caps); | |
606 | } | |
607 | ||
608 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); | |
609 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); | |
610 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, | |
611 | struct ceph_cap *cap); | |
612 | ||
613 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) | |
614 | { | |
615 | int issued; | |
be655596 | 616 | spin_lock(&ci->i_ceph_lock); |
de57606c | 617 | issued = __ceph_caps_issued(ci, NULL); |
be655596 | 618 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
619 | return issued; |
620 | } | |
621 | ||
622 | static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, | |
623 | int touch) | |
624 | { | |
625 | int r; | |
be655596 | 626 | spin_lock(&ci->i_ceph_lock); |
de57606c | 627 | r = __ceph_caps_issued_mask(ci, mask, touch); |
be655596 | 628 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
629 | return r; |
630 | } | |
631 | ||
632 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) | |
633 | { | |
634 | return ci->i_dirty_caps | ci->i_flushing_caps; | |
635 | } | |
f66fd9f0 YZ |
636 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
637 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); | |
638 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, | |
639 | struct ceph_cap_flush **pcf); | |
de57606c | 640 | |
9563f88c YZ |
641 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
642 | struct ceph_cap *ocap, int mask); | |
de57606c SW |
643 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
644 | extern int __ceph_caps_used(struct ceph_inode_info *ci); | |
645 | ||
646 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); | |
647 | ||
648 | /* | |
649 | * wanted, by virtue of open file modes AND cap refs (buffered/cached data) | |
650 | */ | |
651 | static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) | |
652 | { | |
653 | int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); | |
654 | if (w & CEPH_CAP_FILE_BUFFER) | |
655 | w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ | |
656 | return w; | |
657 | } | |
658 | ||
659 | /* what the mds thinks we want */ | |
c1944fed | 660 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
de57606c | 661 | |
37151668 YS |
662 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
663 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); | |
664 | extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); | |
93faca6e | 665 | extern void ceph_reserve_caps(struct ceph_mds_client *mdsc, |
37151668 YS |
666 | struct ceph_cap_reservation *ctx, int need); |
667 | extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, | |
668 | struct ceph_cap_reservation *ctx); | |
3d14c5d2 | 669 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
de57606c | 670 | int *total, int *avail, int *used, |
85ccce43 | 671 | int *reserved, int *min); |
de57606c | 672 | |
de57606c | 673 | |
de57606c SW |
674 | |
675 | /* | |
676 | * we keep buffered readdir results attached to file->private_data | |
677 | */ | |
4918b6d1 | 678 | #define CEPH_F_SYNC 1 |
9cfa1098 | 679 | #define CEPH_F_ATEND 2 |
4918b6d1 | 680 | |
de57606c | 681 | struct ceph_file_info { |
252c6728 SW |
682 | short fmode; /* initialized on open */ |
683 | short flags; /* CEPH_F_* */ | |
de57606c | 684 | |
3bd4d543 YZ |
685 | spinlock_t rw_contexts_lock; |
686 | struct list_head rw_contexts; | |
687 | ||
de57606c SW |
688 | /* readdir: position within the dir */ |
689 | u32 frag; | |
690 | struct ceph_mds_request *last_readdir; | |
de57606c SW |
691 | |
692 | /* readdir: position within a frag */ | |
f0494206 | 693 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
de57606c | 694 | char *last_name; /* last entry in previous chunk */ |
fdd4e158 YZ |
695 | long long dir_release_count; |
696 | long long dir_ordered_count; | |
697 | int readdir_cache_idx; | |
de57606c SW |
698 | |
699 | /* used for -o dirstat read() on directory thing */ | |
700 | char *dir_info; | |
701 | int dir_info_len; | |
702 | }; | |
703 | ||
3bd4d543 YZ |
704 | struct ceph_rw_context { |
705 | struct list_head list; | |
706 | struct task_struct *thread; | |
707 | int caps; | |
708 | }; | |
709 | ||
710 | #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ | |
711 | struct ceph_rw_context _name = { \ | |
712 | .thread = current, \ | |
713 | .caps = _caps, \ | |
714 | } | |
715 | ||
716 | static inline void ceph_add_rw_context(struct ceph_file_info *cf, | |
717 | struct ceph_rw_context *ctx) | |
718 | { | |
719 | spin_lock(&cf->rw_contexts_lock); | |
720 | list_add(&ctx->list, &cf->rw_contexts); | |
721 | spin_unlock(&cf->rw_contexts_lock); | |
722 | } | |
723 | ||
724 | static inline void ceph_del_rw_context(struct ceph_file_info *cf, | |
725 | struct ceph_rw_context *ctx) | |
726 | { | |
727 | spin_lock(&cf->rw_contexts_lock); | |
728 | list_del(&ctx->list); | |
729 | spin_unlock(&cf->rw_contexts_lock); | |
730 | } | |
731 | ||
732 | static inline struct ceph_rw_context* | |
733 | ceph_find_rw_context(struct ceph_file_info *cf) | |
734 | { | |
735 | struct ceph_rw_context *ctx, *found = NULL; | |
736 | spin_lock(&cf->rw_contexts_lock); | |
737 | list_for_each_entry(ctx, &cf->rw_contexts, list) { | |
738 | if (ctx->thread == current) { | |
739 | found = ctx; | |
740 | break; | |
741 | } | |
742 | } | |
743 | spin_unlock(&cf->rw_contexts_lock); | |
744 | return found; | |
745 | } | |
746 | ||
fdd4e158 YZ |
747 | struct ceph_readdir_cache_control { |
748 | struct page *page; | |
749 | struct dentry **dentries; | |
750 | int index; | |
751 | }; | |
de57606c | 752 | |
de57606c SW |
753 | /* |
754 | * A "snap realm" describes a subset of the file hierarchy sharing | |
755 | * the same set of snapshots that apply to it. The realms themselves | |
756 | * are organized into a hierarchy, such that children inherit (some of) | |
757 | * the snapshots of their parents. | |
758 | * | |
759 | * All inodes within the realm that have capabilities are linked into a | |
760 | * per-realm list. | |
761 | */ | |
762 | struct ceph_snap_realm { | |
763 | u64 ino; | |
764 | atomic_t nref; | |
a105f00c SW |
765 | struct rb_node node; |
766 | ||
de57606c SW |
767 | u64 created, seq; |
768 | u64 parent_ino; | |
769 | u64 parent_since; /* snapid when our current parent became so */ | |
770 | ||
771 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ | |
aa711ee3 | 772 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
de57606c | 773 | u64 *snaps; /* snaps specific to this realm */ |
aa711ee3 | 774 | u32 num_snaps; |
de57606c SW |
775 | |
776 | struct ceph_snap_realm *parent; | |
777 | struct list_head children; /* list of child realms */ | |
778 | struct list_head child_item; | |
779 | ||
780 | struct list_head empty_item; /* if i have ref==0 */ | |
781 | ||
ae00d4f3 SW |
782 | struct list_head dirty_item; /* if realm needs new context */ |
783 | ||
de57606c SW |
784 | /* the current set of snaps for this realm */ |
785 | struct ceph_snap_context *cached_context; | |
786 | ||
787 | struct list_head inodes_with_caps; | |
788 | spinlock_t inodes_with_caps_lock; | |
789 | }; | |
790 | ||
3d14c5d2 | 791 | static inline int default_congestion_kb(void) |
de57606c | 792 | { |
3d14c5d2 YS |
793 | int congestion_kb; |
794 | ||
795 | /* | |
796 | * Copied from NFS | |
797 | * | |
798 | * congestion size, scale with available memory. | |
799 | * | |
800 | * 64MB: 8192k | |
801 | * 128MB: 11585k | |
802 | * 256MB: 16384k | |
803 | * 512MB: 23170k | |
804 | * 1GB: 32768k | |
805 | * 2GB: 46340k | |
806 | * 4GB: 65536k | |
807 | * 8GB: 92681k | |
808 | * 16GB: 131072k | |
809 | * | |
810 | * This allows larger machines to have larger/more transfers. | |
811 | * Limit the default to 256M | |
812 | */ | |
813 | congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); | |
814 | if (congestion_kb > 256*1024) | |
815 | congestion_kb = 256*1024; | |
816 | ||
817 | return congestion_kb; | |
de57606c SW |
818 | } |
819 | ||
820 | ||
821 | ||
822 | /* snap.c */ | |
823 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, | |
824 | u64 ino); | |
825 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, | |
826 | struct ceph_snap_realm *realm); | |
827 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, | |
828 | struct ceph_snap_realm *realm); | |
829 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, | |
982d6011 YZ |
830 | void *p, void *e, bool deletion, |
831 | struct ceph_snap_realm **realm_ret); | |
de57606c | 832 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
2600d2dd | 833 | struct ceph_mds_session *session, |
de57606c | 834 | struct ceph_msg *msg); |
fc837c8f | 835 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
de57606c SW |
836 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
837 | struct ceph_cap_snap *capsnap); | |
838 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); | |
839 | ||
840 | /* | |
841 | * a cap_snap is "pending" if it is still awaiting an in-progress | |
842 | * sync write (that may/may not still update size, mtime, etc.). | |
843 | */ | |
844 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) | |
845 | { | |
846 | return !list_empty(&ci->i_cap_snaps) && | |
86056090 YZ |
847 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
848 | ci_item)->writing; | |
de57606c SW |
849 | } |
850 | ||
de57606c SW |
851 | /* inode.c */ |
852 | extern const struct inode_operations ceph_file_iops; | |
853 | ||
854 | extern struct inode *ceph_alloc_inode(struct super_block *sb); | |
855 | extern void ceph_destroy_inode(struct inode *inode); | |
9f12bd11 | 856 | extern int ceph_drop_inode(struct inode *inode); |
de57606c SW |
857 | |
858 | extern struct inode *ceph_get_inode(struct super_block *sb, | |
859 | struct ceph_vino vino); | |
860 | extern struct inode *ceph_get_snapdir(struct inode *parent); | |
861 | extern int ceph_fill_file_size(struct inode *inode, int issued, | |
862 | u32 truncate_seq, u64 truncate_size, u64 size); | |
863 | extern void ceph_fill_file_time(struct inode *inode, int issued, | |
864 | u64 time_warp_seq, struct timespec *ctime, | |
865 | struct timespec *mtime, struct timespec *atime); | |
866 | extern int ceph_fill_trace(struct super_block *sb, | |
f5a03b08 | 867 | struct ceph_mds_request *req); |
de57606c SW |
868 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
869 | struct ceph_mds_session *session); | |
870 | ||
871 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); | |
872 | ||
efb0ca76 | 873 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
b415bf4f | 874 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
3c6f6b79 | 875 | extern void ceph_queue_vmtruncate(struct inode *inode); |
3c6f6b79 SW |
876 | extern void ceph_queue_invalidate(struct inode *inode); |
877 | extern void ceph_queue_writeback(struct inode *inode); | |
0c0e5720 | 878 | extern void ceph_async_iput(struct inode *inode); |
de57606c | 879 | |
01deead0 YZ |
880 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
881 | int mask, bool force); | |
882 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) | |
883 | { | |
884 | return __ceph_do_getattr(inode, NULL, mask, force); | |
885 | } | |
10556cb2 | 886 | extern int ceph_permission(struct inode *inode, int mask); |
a26fecca | 887 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
de57606c | 888 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
a528d35e DH |
889 | extern int ceph_getattr(const struct path *path, struct kstat *stat, |
890 | u32 request_mask, unsigned int flags); | |
de57606c SW |
891 | |
892 | /* xattr.c */ | |
a26fecca | 893 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
7221fe4c | 894 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
de57606c | 895 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
de57606c SW |
896 | extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
897 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); | |
3ce6cd12 AE |
898 | extern void __init ceph_xattr_init(void); |
899 | extern void ceph_xattr_exit(void); | |
b1ee94aa | 900 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
de57606c | 901 | |
315f2408 YZ |
902 | #ifdef CONFIG_SECURITY |
903 | extern bool ceph_security_xattr_deadlock(struct inode *in); | |
904 | extern bool ceph_security_xattr_wanted(struct inode *in); | |
905 | #else | |
906 | static inline bool ceph_security_xattr_deadlock(struct inode *in) | |
907 | { | |
908 | return false; | |
909 | } | |
910 | static inline bool ceph_security_xattr_wanted(struct inode *in) | |
911 | { | |
912 | return false; | |
913 | } | |
914 | #endif | |
915 | ||
7221fe4c | 916 | /* acl.c */ |
b1ee94aa YZ |
917 | struct ceph_acls_info { |
918 | void *default_acl; | |
919 | void *acl; | |
920 | struct ceph_pagelist *pagelist; | |
921 | }; | |
7221fe4c GZ |
922 | |
923 | #ifdef CONFIG_CEPH_FS_POSIX_ACL | |
924 | ||
925 | struct posix_acl *ceph_get_acl(struct inode *, int); | |
72466d0b | 926 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
b1ee94aa YZ |
927 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
928 | struct ceph_acls_info *info); | |
929 | void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); | |
930 | void ceph_release_acls_info(struct ceph_acls_info *info); | |
c969d9bf GZ |
931 | |
932 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
933 | { | |
934 | forget_all_cached_acls(inode); | |
935 | } | |
7221fe4c GZ |
936 | |
937 | #else | |
938 | ||
939 | #define ceph_get_acl NULL | |
72466d0b | 940 | #define ceph_set_acl NULL |
7221fe4c | 941 | |
b1ee94aa YZ |
942 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
943 | struct ceph_acls_info *info) | |
7221fe4c GZ |
944 | { |
945 | return 0; | |
946 | } | |
b1ee94aa YZ |
947 | static inline void ceph_init_inode_acls(struct inode *inode, |
948 | struct ceph_acls_info *info) | |
949 | { | |
950 | } | |
951 | static inline void ceph_release_acls_info(struct ceph_acls_info *info) | |
952 | { | |
953 | } | |
7221fe4c GZ |
954 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
955 | { | |
956 | return 0; | |
957 | } | |
958 | ||
959 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
960 | { | |
961 | } | |
962 | ||
963 | #endif | |
964 | ||
de57606c SW |
965 | /* caps.c */ |
966 | extern const char *ceph_cap_string(int c); | |
967 | extern void ceph_handle_caps(struct ceph_mds_session *session, | |
968 | struct ceph_msg *msg); | |
d9df2783 YZ |
969 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
970 | struct ceph_cap_reservation *ctx); | |
971 | extern void ceph_add_cap(struct inode *inode, | |
972 | struct ceph_mds_session *session, u64 cap_id, | |
973 | int fmode, unsigned issued, unsigned wanted, | |
974 | unsigned cap, unsigned seq, u64 realmino, int flags, | |
975 | struct ceph_cap **new_cap); | |
a096b09a | 976 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
1de82206 | 977 | extern void __ceph_remove_caps(struct inode* inode); |
37151668 YS |
978 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
979 | struct ceph_cap *cap); | |
9215aeea | 980 | extern int ceph_is_any_caps(struct inode *inode); |
de57606c | 981 | |
f1a3d572 | 982 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
02c24a82 JB |
983 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
984 | int datasync); | |
e548e9b9 YZ |
985 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
986 | struct ceph_mds_session *session); | |
de57606c SW |
987 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
988 | struct ceph_mds_session *session); | |
2bc50259 GF |
989 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
990 | int mds); | |
de57606c SW |
991 | extern int ceph_get_cap_mds(struct inode *inode); |
992 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); | |
993 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); | |
994 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, | |
995 | struct ceph_snap_context *snapc); | |
ed9b430c YZ |
996 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
997 | struct ceph_mds_session **psession); | |
efb0ca76 | 998 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
de57606c SW |
999 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
1000 | struct ceph_mds_session *session); | |
afcdaea3 SW |
1001 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
1002 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); | |
de57606c SW |
1003 | |
1004 | extern int ceph_encode_inode_release(void **p, struct inode *inode, | |
1005 | int mds, int drop, int unless, int force); | |
1006 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, | |
ca6c8ae0 | 1007 | struct inode *dir, |
de57606c SW |
1008 | int mds, int drop, int unless); |
1009 | ||
1010 | extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, | |
31c542a1 | 1011 | loff_t endoff, int *got, struct page **pinned_page); |
2b1ac852 YZ |
1012 | extern int ceph_try_get_caps(struct ceph_inode_info *ci, |
1013 | int need, int want, int *got); | |
de57606c SW |
1014 | |
1015 | /* for counting open files by mode */ | |
774a6a11 | 1016 | extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); |
de57606c SW |
1017 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); |
1018 | ||
1019 | /* addr.c */ | |
1020 | extern const struct address_space_operations ceph_aops; | |
1021 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); | |
10183a69 YZ |
1022 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
1023 | extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); | |
1024 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); | |
de57606c SW |
1025 | |
1026 | /* file.c */ | |
1027 | extern const struct file_operations ceph_file_fops; | |
9e0eb85d | 1028 | |
77310320 | 1029 | extern int ceph_renew_caps(struct inode *inode); |
de57606c | 1030 | extern int ceph_open(struct inode *inode, struct file *file); |
5ef50c3b SW |
1031 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
1032 | struct file *file, unsigned flags, umode_t mode, | |
1033 | int *opened); | |
de57606c | 1034 | extern int ceph_release(struct inode *inode, struct file *filp); |
31c542a1 YZ |
1035 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
1036 | char *data, size_t len); | |
55f2a045 | 1037 | |
de57606c SW |
1038 | /* dir.c */ |
1039 | extern const struct file_operations ceph_dir_fops; | |
38c48b5f | 1040 | extern const struct file_operations ceph_snapdir_fops; |
de57606c | 1041 | extern const struct inode_operations ceph_dir_iops; |
38c48b5f | 1042 | extern const struct inode_operations ceph_snapdir_iops; |
18fc8abd | 1043 | extern const struct dentry_operations ceph_dentry_ops; |
de57606c | 1044 | |
f3c4ebe6 | 1045 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
de57606c | 1046 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
468640e3 SW |
1047 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
1048 | struct dentry *dentry, int err); | |
de57606c SW |
1049 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
1050 | struct dentry *dentry, int err); | |
1051 | ||
1052 | extern void ceph_dentry_lru_add(struct dentry *dn); | |
1053 | extern void ceph_dentry_lru_touch(struct dentry *dn); | |
1054 | extern void ceph_dentry_lru_del(struct dentry *dn); | |
81a6cf2d | 1055 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
e5f86dc3 | 1056 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
fdd4e158 | 1057 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
de57606c | 1058 | |
de57606c SW |
1059 | /* ioctl.c */ |
1060 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); | |
1061 | ||
1062 | /* export.c */ | |
1063 | extern const struct export_operations ceph_export_ops; | |
1064 | ||
40819f6f | 1065 | /* locks.c */ |
eb13e832 | 1066 | extern __init void ceph_flock_init(void); |
40819f6f GF |
1067 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
1068 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); | |
1069 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); | |
39be95e9 JS |
1070 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
1071 | struct ceph_filelock *flocks, | |
1072 | int num_fcntl_locks, | |
1073 | int num_flock_locks); | |
1074 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, | |
1075 | struct ceph_pagelist *pagelist, | |
1076 | int num_fcntl_locks, int num_flock_locks); | |
40819f6f | 1077 | |
3d14c5d2 YS |
1078 | /* debugfs.c */ |
1079 | extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); | |
1080 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); | |
1081 | ||
7370e8a4 | 1082 | /* quota.c */ |
60147660 LH |
1083 | static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) |
1084 | { | |
1085 | return ci->i_max_files || ci->i_max_bytes; | |
1086 | } | |
1087 | ||
1088 | extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); | |
1089 | ||
1090 | static inline void __ceph_update_quota(struct ceph_inode_info *ci, | |
1091 | u64 max_bytes, u64 max_files) | |
1092 | { | |
1093 | bool had_quota, has_quota; | |
1094 | had_quota = __ceph_has_any_quota(ci); | |
1095 | ci->i_max_bytes = max_bytes; | |
1096 | ci->i_max_files = max_files; | |
1097 | has_quota = __ceph_has_any_quota(ci); | |
1098 | ||
1099 | if (had_quota != has_quota) | |
1100 | ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); | |
1101 | } | |
1102 | ||
7370e8a4 LH |
1103 | extern void ceph_handle_quota(struct ceph_mds_client *mdsc, |
1104 | struct ceph_mds_session *session, | |
1105 | struct ceph_msg *msg); | |
6aa007ee | 1106 | extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); |
68a16a23 | 1107 | extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); |
0cb65d53 LH |
1108 | extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, |
1109 | loff_t newlen); | |
763b0c3d LH |
1110 | extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, |
1111 | loff_t newlen); | |
7370e8a4 | 1112 | |
de57606c | 1113 | #endif /* _FS_CEPH_SUPER_H */ |