]>
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> |
668959a5 | 19 | #include <linux/security.h> |
de57606c | 20 | |
3d14c5d2 | 21 | #include <linux/ceph/libceph.h> |
de57606c | 22 | |
99ccbd22 MT |
23 | #ifdef CONFIG_CEPH_FSCACHE |
24 | #include <linux/fscache.h> | |
25 | #endif | |
26 | ||
de57606c SW |
27 | /* f_type in struct statfs */ |
28 | #define CEPH_SUPER_MAGIC 0x00c36400 | |
29 | ||
30 | /* large granularity for statfs utilization stats to facilitate | |
31 | * large volume sizes on 32-bit machines. */ | |
92a49fb0 | 32 | #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ |
de57606c SW |
33 | #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) |
34 | ||
131d7eb4 | 35 | #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blacklisted */ |
3d14c5d2 YS |
36 | #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ |
37 | #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ | |
38 | #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ | |
ad1fee96 | 39 | #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ |
a40dc6cc | 40 | #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ |
99ccbd22 | 41 | #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ |
10183a69 | 42 | #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ |
e9e427f0 | 43 | #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ |
9122eed5 | 44 | #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ |
ea4cdc54 | 45 | #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ |
2ccb4546 | 46 | #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ |
6a259382 | 47 | |
6f9718fe LH |
48 | #define CEPH_MOUNT_OPT_DEFAULT \ |
49 | (CEPH_MOUNT_OPT_DCACHE | \ | |
50 | CEPH_MOUNT_OPT_NOCOPYFROM) | |
de57606c | 51 | |
3d14c5d2 | 52 | #define ceph_set_mount_opt(fsc, opt) \ |
2ccb4546 JL |
53 | (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt |
54 | #define ceph_clear_mount_opt(fsc, opt) \ | |
55 | (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt | |
3d14c5d2 YS |
56 | #define ceph_test_mount_opt(fsc, opt) \ |
57 | (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) | |
de57606c | 58 | |
aa187926 YZ |
59 | /* max size of osd read request, limited by libceph */ |
60 | #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN | |
95cca2b4 YZ |
61 | /* osd has a configurable limitaion of max write size. |
62 | * CEPH_MSG_MAX_DATA_LEN should be small enough. */ | |
63 | #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN | |
7c94ba27 | 64 | #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ |
3d14c5d2 YS |
65 | #define CEPH_MAX_READDIR_DEFAULT 1024 |
66 | #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) | |
67 | #define CEPH_SNAPDIRNAME_DEFAULT ".snap" | |
de57606c | 68 | |
4214fb15 YZ |
69 | /* |
70 | * Delay telling the MDS we no longer want caps, in case we reopen | |
71 | * the file. Delay a minimum amount of time, even if we send a cap | |
72 | * message for some other reason. Otherwise, take the oppotunity to | |
73 | * update the mds to avoid sending another message later. | |
74 | */ | |
75 | #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ | |
76 | #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ | |
77 | ||
3d14c5d2 | 78 | struct ceph_mount_options { |
ad8c28a9 | 79 | unsigned int flags; |
3d14c5d2 | 80 | |
ad8c28a9 JL |
81 | unsigned int wsize; /* max write size */ |
82 | unsigned int rsize; /* max read size */ | |
83 | unsigned int rasize; /* max readahead */ | |
84 | unsigned int congestion_kb; /* max writeback in flight */ | |
85 | unsigned int caps_wanted_delay_min, caps_wanted_delay_max; | |
fe33032d | 86 | int caps_max; |
ad8c28a9 JL |
87 | unsigned int max_readdir; /* max readdir result (entries) */ |
88 | unsigned int max_readdir_bytes; /* max readdir result (bytes) */ | |
de57606c | 89 | |
3d14c5d2 YS |
90 | /* |
91 | * everything above this point can be memcmp'd; everything below | |
92 | * is handled in compare_mount_options() | |
93 | */ | |
de57606c | 94 | |
3d14c5d2 | 95 | char *snapdir_name; /* default ".snap" */ |
430afbad | 96 | char *mds_namespace; /* default NULL */ |
b27a939e | 97 | char *server_path; /* default NULL (means "/") */ |
1d8f8360 | 98 | char *fscache_uniq; /* default NULL */ |
3d14c5d2 | 99 | }; |
de57606c | 100 | |
3d14c5d2 | 101 | struct ceph_fs_client { |
de57606c SW |
102 | struct super_block *sb; |
103 | ||
18f473b3 XL |
104 | struct list_head metric_wakeup; |
105 | ||
3d14c5d2 YS |
106 | struct ceph_mount_options *mount_options; |
107 | struct ceph_client *client; | |
de57606c | 108 | |
3d14c5d2 | 109 | unsigned long mount_state; |
81f148a9 | 110 | |
131d7eb4 YZ |
111 | unsigned long last_auto_reconnect; |
112 | bool blacklisted; | |
113 | ||
78beb0ff LH |
114 | bool have_copy_from2; |
115 | ||
81f148a9 | 116 | u32 filp_gen; |
719784ba | 117 | loff_t max_file_size; |
85ccce43 | 118 | |
3d14c5d2 | 119 | struct ceph_mds_client *mdsc; |
de57606c SW |
120 | |
121 | /* writeback */ | |
122 | mempool_t *wb_pagevec_pool; | |
e3ec8d68 YZ |
123 | atomic_long_t writeback_count; |
124 | ||
1cf89a8d | 125 | struct workqueue_struct *inode_wq; |
e3ec8d68 | 126 | struct workqueue_struct *cap_wq; |
de57606c | 127 | |
0743304d | 128 | #ifdef CONFIG_DEBUG_FS |
3d14c5d2 | 129 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
2baba250 | 130 | struct dentry *debugfs_congestion_kb; |
06edf046 | 131 | struct dentry *debugfs_bdi; |
3d14c5d2 | 132 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
f9009efa | 133 | struct dentry *debugfs_metric; |
14ed9703 | 134 | struct dentry *debugfs_mds_sessions; |
0743304d | 135 | #endif |
99ccbd22 MT |
136 | |
137 | #ifdef CONFIG_CEPH_FSCACHE | |
138 | struct fscache_cookie *fscache; | |
99ccbd22 | 139 | #endif |
de57606c SW |
140 | }; |
141 | ||
3d14c5d2 | 142 | |
de57606c SW |
143 | /* |
144 | * File i/o capability. This tracks shared state with the metadata | |
145 | * server that allows us to cache or writeback attributes or to read | |
146 | * and write data. For any given inode, we should have one or more | |
147 | * capabilities, one issued by each metadata server, and our | |
148 | * cumulative access is the OR of all issued capabilities. | |
149 | * | |
150 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds | |
151 | * session capability lists. | |
152 | */ | |
153 | struct ceph_cap { | |
154 | struct ceph_inode_info *ci; | |
155 | struct rb_node ci_node; /* per-ci cap tree */ | |
156 | struct ceph_mds_session *session; | |
157 | struct list_head session_caps; /* per-session caplist */ | |
de57606c | 158 | u64 cap_id; /* unique cap id (mds provided) */ |
745a8e3b YZ |
159 | union { |
160 | /* in-use caps */ | |
161 | struct { | |
162 | int issued; /* latest, from the mds */ | |
163 | int implemented; /* implemented superset of | |
164 | issued (for revocation) */ | |
165 | int mds, mds_wanted; | |
166 | }; | |
167 | /* caps to release */ | |
168 | struct { | |
169 | u64 cap_ino; | |
170 | int queue_release; | |
171 | }; | |
172 | }; | |
685f9a5d SW |
173 | u32 seq, issue_seq, mseq; |
174 | u32 cap_gen; /* active/stale cycle */ | |
de57606c SW |
175 | unsigned long last_used; |
176 | struct list_head caps_item; | |
177 | }; | |
178 | ||
a0d93e32 YZ |
179 | #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */ |
180 | #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */ | |
181 | #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */ | |
de57606c | 182 | |
0e294387 YZ |
183 | struct ceph_cap_flush { |
184 | u64 tid; | |
185 | int caps; /* 0 means capsnap */ | |
c8799fc4 | 186 | bool wake; /* wake up flush waiters when finish ? */ |
0e294387 YZ |
187 | struct list_head g_list; // global |
188 | struct list_head i_list; // per inode | |
189 | }; | |
190 | ||
de57606c SW |
191 | /* |
192 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, | |
193 | * we first complete any in-process sync writes and writeback any dirty | |
194 | * data before flushing the snapped state (tracked here) back to the MDS. | |
195 | */ | |
196 | struct ceph_cap_snap { | |
805692d0 | 197 | refcount_t nref; |
0e294387 | 198 | struct list_head ci_item; |
de57606c | 199 | |
0e294387 YZ |
200 | struct ceph_cap_flush cap_flush; |
201 | ||
202 | u64 follows; | |
de57606c SW |
203 | int issued, dirty; |
204 | struct ceph_snap_context *context; | |
205 | ||
5706b27d | 206 | umode_t mode; |
05cb11c1 EB |
207 | kuid_t uid; |
208 | kgid_t gid; | |
de57606c | 209 | |
4a625be4 | 210 | struct ceph_buffer *xattr_blob; |
de57606c SW |
211 | u64 xattr_version; |
212 | ||
213 | u64 size; | |
176c77c9 | 214 | u64 change_attr; |
ec62b894 | 215 | struct timespec64 mtime, atime, ctime, btime; |
de57606c | 216 | u64 time_warp_seq; |
5f743e45 YZ |
217 | u64 truncate_size; |
218 | u32 truncate_seq; | |
de57606c SW |
219 | int writing; /* a sync write is still in progress */ |
220 | int dirty_pages; /* dirty pages awaiting writeback */ | |
e20d258d | 221 | bool inline_data; |
86056090 | 222 | bool need_flush; |
de57606c SW |
223 | }; |
224 | ||
225 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) | |
226 | { | |
805692d0 | 227 | if (refcount_dec_and_test(&capsnap->nref)) { |
4a625be4 SW |
228 | if (capsnap->xattr_blob) |
229 | ceph_buffer_put(capsnap->xattr_blob); | |
de57606c | 230 | kfree(capsnap); |
4a625be4 | 231 | } |
de57606c SW |
232 | } |
233 | ||
234 | /* | |
235 | * The frag tree describes how a directory is fragmented, potentially across | |
236 | * multiple metadata servers. It is also used to indicate points where | |
237 | * metadata authority is delegated, and whether/where metadata is replicated. | |
238 | * | |
239 | * A _leaf_ frag will be present in the i_fragtree IFF there is | |
240 | * delegation info. That is, if mds >= 0 || ndist > 0. | |
241 | */ | |
242 | #define CEPH_MAX_DIRFRAG_REP 4 | |
243 | ||
244 | struct ceph_inode_frag { | |
245 | struct rb_node node; | |
246 | ||
247 | /* fragtree state */ | |
248 | u32 frag; | |
249 | int split_by; /* i.e. 2^(split_by) children */ | |
250 | ||
251 | /* delegation and replication info */ | |
252 | int mds; /* -1 if same authority as parent */ | |
253 | int ndist; /* >0 if replicated */ | |
254 | int dist[CEPH_MAX_DIRFRAG_REP]; | |
255 | }; | |
256 | ||
257 | /* | |
258 | * We cache inode xattrs as an encoded blob until they are first used, | |
259 | * at which point we parse them into an rbtree. | |
260 | */ | |
261 | struct ceph_inode_xattr { | |
262 | struct rb_node node; | |
263 | ||
264 | const char *name; | |
265 | int name_len; | |
266 | const char *val; | |
267 | int val_len; | |
268 | int dirty; | |
269 | ||
270 | int should_free_name; | |
271 | int should_free_val; | |
272 | }; | |
273 | ||
3d14c5d2 YS |
274 | /* |
275 | * Ceph dentry state | |
276 | */ | |
277 | struct ceph_dentry_info { | |
37c4efc1 | 278 | struct dentry *dentry; |
3d14c5d2 | 279 | struct ceph_mds_session *lease_session; |
37c4efc1 YZ |
280 | struct list_head lease_list; |
281 | unsigned flags; | |
97aeb6bf YZ |
282 | int lease_shared_gen; |
283 | u32 lease_gen; | |
3d14c5d2 YS |
284 | u32 lease_seq; |
285 | unsigned long lease_renew_after, lease_renew_from; | |
9b16f03c | 286 | unsigned long time; |
3d14c5d2 YS |
287 | u64 offset; |
288 | }; | |
289 | ||
37c4efc1 YZ |
290 | #define CEPH_DENTRY_REFERENCED 1 |
291 | #define CEPH_DENTRY_LEASE_LIST 2 | |
292 | #define CEPH_DENTRY_SHRINK_LIST 4 | |
f5e17aed | 293 | #define CEPH_DENTRY_PRIMARY_LINK 8 |
37c4efc1 | 294 | |
de57606c SW |
295 | struct ceph_inode_xattrs_info { |
296 | /* | |
297 | * (still encoded) xattr blob. we avoid the overhead of parsing | |
298 | * this until someone actually calls getxattr, etc. | |
299 | * | |
300 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == | |
301 | * NULL means we don't know. | |
302 | */ | |
303 | struct ceph_buffer *blob, *prealloc_blob; | |
304 | ||
305 | struct rb_root index; | |
306 | bool dirty; | |
307 | int count; | |
308 | int names_size; | |
309 | int vals_size; | |
310 | u64 version, index_version; | |
311 | }; | |
312 | ||
313 | /* | |
314 | * Ceph inode. | |
315 | */ | |
de57606c SW |
316 | struct ceph_inode_info { |
317 | struct ceph_vino i_vino; /* ceph ino + snap */ | |
318 | ||
be655596 SW |
319 | spinlock_t i_ceph_lock; |
320 | ||
de57606c | 321 | u64 i_version; |
31c542a1 | 322 | u64 i_inline_version; |
de57606c SW |
323 | u32 i_time_warp_seq; |
324 | ||
891f3f5a | 325 | unsigned long i_ceph_flags; |
fdd4e158 YZ |
326 | atomic64_t i_release_count; |
327 | atomic64_t i_ordered_count; | |
328 | atomic64_t i_complete_seq[2]; | |
de57606c | 329 | |
6c0f3af7 | 330 | struct ceph_dir_layout i_dir_layout; |
de57606c | 331 | struct ceph_file_layout i_layout; |
785892fe | 332 | struct ceph_file_layout i_cached_layout; // for async creates |
de57606c SW |
333 | char *i_symlink; |
334 | ||
335 | /* for dirs */ | |
9bbeab41 | 336 | struct timespec64 i_rctime; |
de57606c SW |
337 | u64 i_rbytes, i_rfiles, i_rsubdirs; |
338 | u64 i_files, i_subdirs; | |
de57606c | 339 | |
fb18a575 LH |
340 | /* quotas */ |
341 | u64 i_max_bytes, i_max_files; | |
342 | ||
08796873 YZ |
343 | s32 i_dir_pin; |
344 | ||
de57606c | 345 | struct rb_root i_fragtree; |
1b1bc16d | 346 | int i_fragtree_nsplits; |
de57606c SW |
347 | struct mutex i_fragtree_mutex; |
348 | ||
349 | struct ceph_inode_xattrs_info i_xattrs; | |
350 | ||
be655596 | 351 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
de57606c SW |
352 | * s_mutex. */ |
353 | struct rb_root i_caps; /* cap list */ | |
354 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ | |
355 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ | |
1cf03a68 JL |
356 | |
357 | /* | |
358 | * Link to the the auth cap's session's s_cap_dirty list. s_cap_dirty | |
359 | * is protected by the mdsc->cap_dirty_lock, but each individual item | |
360 | * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty | |
361 | * requires the mdsc->cap_dirty_lock. List presence for an item can | |
362 | * be tested under the i_ceph_lock. Changing anything requires both. | |
363 | */ | |
364 | struct list_head i_dirty_item; | |
365 | ||
366 | /* | |
829ad4db JL |
367 | * Link to session's s_cap_flushing list. Protected in a similar |
368 | * fashion to i_dirty_item, but also by the s_mutex for changes. The | |
369 | * s_cap_flushing list can be walked while holding either the s_mutex | |
370 | * or msdc->cap_dirty_lock. List presence can also be checked while | |
371 | * holding the i_ceph_lock for this inode. | |
1cf03a68 JL |
372 | */ |
373 | struct list_head i_flushing_item; | |
374 | ||
de57606c SW |
375 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
376 | * overlapping, pipelined cap flushes to the mds. we can probably | |
377 | * reduce the tid to 8 bits if we're concerned about inode size. */ | |
f66fd9f0 | 378 | struct ceph_cap_flush *i_prealloc_cap_flush; |
e4500b5e | 379 | struct list_head i_cap_flush_list; |
de57606c | 380 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
de57606c SW |
381 | unsigned long i_hold_caps_max; /* jiffies */ |
382 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ | |
de57606c SW |
383 | struct ceph_cap_reservation i_cap_migration_resv; |
384 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ | |
7d8cb26d SW |
385 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
386 | dirty|flushing caps */ | |
de57606c SW |
387 | unsigned i_snap_caps; /* cap bits for snapped files */ |
388 | ||
719a2514 YZ |
389 | unsigned long i_last_rd; |
390 | unsigned long i_last_wr; | |
774a6a11 | 391 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
de57606c | 392 | |
b0d7c223 | 393 | struct mutex i_truncate_mutex; |
de57606c SW |
394 | u32 i_truncate_seq; /* last truncate to smaller size */ |
395 | u64 i_truncate_size; /* and the size we last truncated down to */ | |
396 | int i_truncate_pending; /* still need to call vmtruncate */ | |
397 | ||
398 | u64 i_max_size; /* max file size authorized by mds */ | |
399 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ | |
400 | u64 i_wanted_max_size; /* offset we'd like to write too */ | |
401 | u64 i_requested_max_size; /* max_size we've requested */ | |
402 | ||
403 | /* held references to caps */ | |
404 | int i_pin_ref; | |
f85122af | 405 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref; |
de57606c | 406 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
89aa5930 | 407 | atomic_t i_filelock_ref; |
97aeb6bf | 408 | atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ |
cd045cb4 | 409 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
de57606c SW |
410 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
411 | ||
de57606c | 412 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
68cd5b4b | 413 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
de57606c SW |
414 | spinlock_t i_unsafe_lock; |
415 | ||
75c9627e YZ |
416 | union { |
417 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ | |
418 | struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ | |
419 | }; | |
de57606c SW |
420 | int i_snap_realm_counter; /* snap realm (if caps) */ |
421 | struct list_head i_snap_realm_item; | |
422 | struct list_head i_snap_flush_item; | |
245ce991 | 423 | struct timespec64 i_btime; |
193e7b37 | 424 | struct timespec64 i_snap_btime; |
de57606c | 425 | |
1cf89a8d YZ |
426 | struct work_struct i_work; |
427 | unsigned long i_work_mask; | |
de57606c | 428 | |
99ccbd22 MT |
429 | #ifdef CONFIG_CEPH_FSCACHE |
430 | struct fscache_cookie *fscache; | |
f7f7e7a0 | 431 | u32 i_fscache_gen; |
99ccbd22 | 432 | #endif |
f4b97866 YZ |
433 | errseq_t i_meta_err; |
434 | ||
de57606c SW |
435 | struct inode vfs_inode; /* at end */ |
436 | }; | |
437 | ||
721d5c13 JL |
438 | static inline struct ceph_inode_info * |
439 | ceph_inode(const struct inode *inode) | |
de57606c | 440 | { |
fbbccec9 | 441 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
de57606c SW |
442 | } |
443 | ||
721d5c13 JL |
444 | static inline struct ceph_fs_client * |
445 | ceph_inode_to_client(const struct inode *inode) | |
ad1fee96 YS |
446 | { |
447 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; | |
448 | } | |
449 | ||
721d5c13 JL |
450 | static inline struct ceph_fs_client * |
451 | ceph_sb_to_client(const struct super_block *sb) | |
ad1fee96 YS |
452 | { |
453 | return (struct ceph_fs_client *)sb->s_fs_info; | |
454 | } | |
455 | ||
721d5c13 JL |
456 | static inline struct ceph_vino |
457 | ceph_vino(const struct inode *inode) | |
3d14c5d2 YS |
458 | { |
459 | return ceph_inode(inode)->i_vino; | |
460 | } | |
461 | ||
462 | /* | |
463 | * ino_t is <64 bits on many architectures, blech. | |
464 | * | |
ad1fee96 YS |
465 | * i_ino (kernel inode) st_ino (userspace) |
466 | * i386 32 32 | |
467 | * x86_64+ino32 64 32 | |
468 | * x86_64 64 64 | |
469 | */ | |
3310f754 | 470 | static inline u32 ceph_ino_to_ino32(__u64 vino) |
ad1fee96 | 471 | { |
3310f754 AO |
472 | u32 ino = vino & 0xffffffff; |
473 | ino ^= vino >> 32; | |
ad1fee96 | 474 | if (!ino) |
a661fc56 | 475 | ino = 2; |
ad1fee96 YS |
476 | return ino; |
477 | } | |
478 | ||
479 | /* | |
480 | * kernel i_ino value | |
3d14c5d2 YS |
481 | */ |
482 | static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) | |
483 | { | |
3d14c5d2 | 484 | #if BITS_PER_LONG == 32 |
3310f754 AO |
485 | return ceph_ino_to_ino32(vino.ino); |
486 | #else | |
487 | return (ino_t)vino.ino; | |
3d14c5d2 | 488 | #endif |
3d14c5d2 YS |
489 | } |
490 | ||
ad1fee96 YS |
491 | /* |
492 | * user-visible ino (stat, filldir) | |
493 | */ | |
494 | #if BITS_PER_LONG == 32 | |
495 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
496 | { | |
497 | return ino; | |
498 | } | |
499 | #else | |
500 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
501 | { | |
502 | if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) | |
503 | ino = ceph_ino_to_ino32(ino); | |
504 | return ino; | |
505 | } | |
506 | #endif | |
507 | ||
508 | ||
3d14c5d2 YS |
509 | /* for printf-style formatting */ |
510 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap | |
511 | ||
512 | static inline u64 ceph_ino(struct inode *inode) | |
513 | { | |
514 | return ceph_inode(inode)->i_vino.ino; | |
515 | } | |
516 | static inline u64 ceph_snap(struct inode *inode) | |
517 | { | |
518 | return ceph_inode(inode)->i_vino.snap; | |
519 | } | |
520 | ||
521 | static inline int ceph_ino_compare(struct inode *inode, void *data) | |
522 | { | |
523 | struct ceph_vino *pvino = (struct ceph_vino *)data; | |
524 | struct ceph_inode_info *ci = ceph_inode(inode); | |
525 | return ci->i_vino.ino == pvino->ino && | |
526 | ci->i_vino.snap == pvino->snap; | |
527 | } | |
528 | ||
529 | static inline struct inode *ceph_find_inode(struct super_block *sb, | |
530 | struct ceph_vino vino) | |
531 | { | |
532 | ino_t t = ceph_vino_to_ino(vino); | |
533 | return ilookup5(sb, t, ceph_ino_compare, &vino); | |
534 | } | |
535 | ||
536 | ||
537 | /* | |
538 | * Ceph inode. | |
539 | */ | |
10183a69 | 540 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
10183a69 | 541 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ |
daca8bda JL |
542 | #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */ |
543 | #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */ | |
544 | #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */ | |
545 | #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */ | |
c0e385b1 YZ |
546 | #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */ |
547 | #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */ | |
548 | #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */ | |
549 | #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */ | |
550 | #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */ | |
551 | #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */ | |
891f3f5a | 552 | #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT) |
26544c62 | 553 | |
1cf89a8d YZ |
554 | /* |
555 | * Masks of ceph inode work. | |
556 | */ | |
557 | #define CEPH_I_WORK_WRITEBACK 0 /* writeback */ | |
558 | #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */ | |
559 | #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */ | |
560 | ||
26544c62 JL |
561 | /* |
562 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode | |
563 | * and then clear it when they start succeeding. Note that we do a lockless | |
564 | * check first, and only take the lock if it looks like it needs to be changed. | |
565 | * The write submission code just takes this as a hint, so we're not too | |
566 | * worried if a few slip through in either direction. | |
567 | */ | |
568 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) | |
569 | { | |
570 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { | |
571 | spin_lock(&ci->i_ceph_lock); | |
572 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; | |
573 | spin_unlock(&ci->i_ceph_lock); | |
574 | } | |
575 | } | |
576 | ||
577 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) | |
578 | { | |
579 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { | |
580 | spin_lock(&ci->i_ceph_lock); | |
581 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; | |
582 | spin_unlock(&ci->i_ceph_lock); | |
583 | } | |
584 | } | |
3d14c5d2 | 585 | |
2f276c51 | 586 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
fdd4e158 YZ |
587 | long long release_count, |
588 | long long ordered_count) | |
de57606c | 589 | { |
74960773 AP |
590 | /* |
591 | * Makes sure operations that setup readdir cache (update page | |
592 | * cache and i_size) are strongly ordered w.r.t. the following | |
593 | * atomic64_set() operations. | |
594 | */ | |
595 | smp_mb(); | |
fdd4e158 YZ |
596 | atomic64_set(&ci->i_complete_seq[0], release_count); |
597 | atomic64_set(&ci->i_complete_seq[1], ordered_count); | |
de57606c SW |
598 | } |
599 | ||
2f276c51 | 600 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
de57606c | 601 | { |
fdd4e158 YZ |
602 | atomic64_inc(&ci->i_release_count); |
603 | } | |
604 | ||
605 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) | |
606 | { | |
607 | atomic64_inc(&ci->i_ordered_count); | |
2f276c51 | 608 | } |
de57606c | 609 | |
2f276c51 YZ |
610 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
611 | { | |
fdd4e158 YZ |
612 | return atomic64_read(&ci->i_complete_seq[0]) == |
613 | atomic64_read(&ci->i_release_count); | |
de57606c SW |
614 | } |
615 | ||
70db4f36 YZ |
616 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
617 | { | |
fdd4e158 YZ |
618 | return atomic64_read(&ci->i_complete_seq[0]) == |
619 | atomic64_read(&ci->i_release_count) && | |
620 | atomic64_read(&ci->i_complete_seq[1]) == | |
621 | atomic64_read(&ci->i_ordered_count); | |
70db4f36 YZ |
622 | } |
623 | ||
2f276c51 | 624 | static inline void ceph_dir_clear_complete(struct inode *inode) |
de57606c | 625 | { |
2f276c51 YZ |
626 | __ceph_dir_clear_complete(ceph_inode(inode)); |
627 | } | |
de57606c | 628 | |
70db4f36 | 629 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
2f276c51 | 630 | { |
fdd4e158 | 631 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
de57606c SW |
632 | } |
633 | ||
70db4f36 YZ |
634 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
635 | { | |
fdd4e158 YZ |
636 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
637 | smp_rmb(); | |
70db4f36 YZ |
638 | return ret; |
639 | } | |
de57606c SW |
640 | |
641 | /* find a specific frag @f */ | |
642 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, | |
643 | u32 f); | |
644 | ||
645 | /* | |
646 | * choose fragment for value @v. copy frag content to pfrag, if leaf | |
647 | * exists | |
648 | */ | |
649 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, | |
650 | struct ceph_inode_frag *pfrag, | |
651 | int *found); | |
652 | ||
1e9c2eb6 | 653 | static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) |
de57606c SW |
654 | { |
655 | return (struct ceph_dentry_info *)dentry->d_fsdata; | |
656 | } | |
657 | ||
de57606c SW |
658 | /* |
659 | * caps helpers | |
660 | */ | |
661 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) | |
662 | { | |
663 | return !RB_EMPTY_ROOT(&ci->i_caps); | |
664 | } | |
665 | ||
666 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); | |
667 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); | |
1af16d54 XL |
668 | extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask, |
669 | int t); | |
de57606c SW |
670 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, |
671 | struct ceph_cap *cap); | |
672 | ||
673 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) | |
674 | { | |
675 | int issued; | |
be655596 | 676 | spin_lock(&ci->i_ceph_lock); |
de57606c | 677 | issued = __ceph_caps_issued(ci, NULL); |
be655596 | 678 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
679 | return issued; |
680 | } | |
681 | ||
1af16d54 XL |
682 | static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, |
683 | int mask, int touch) | |
de57606c SW |
684 | { |
685 | int r; | |
be655596 | 686 | spin_lock(&ci->i_ceph_lock); |
1af16d54 | 687 | r = __ceph_caps_issued_mask_metric(ci, mask, touch); |
be655596 | 688 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
689 | return r; |
690 | } | |
691 | ||
692 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) | |
693 | { | |
694 | return ci->i_dirty_caps | ci->i_flushing_caps; | |
695 | } | |
f66fd9f0 YZ |
696 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
697 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); | |
698 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, | |
699 | struct ceph_cap_flush **pcf); | |
de57606c | 700 | |
9563f88c YZ |
701 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
702 | struct ceph_cap *ocap, int mask); | |
de57606c SW |
703 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
704 | extern int __ceph_caps_used(struct ceph_inode_info *ci); | |
705 | ||
719a2514 YZ |
706 | static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci) |
707 | { | |
708 | return ci->i_nr_by_mode[0]; | |
709 | } | |
de57606c | 710 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); |
525d15e8 | 711 | extern int __ceph_caps_wanted(struct ceph_inode_info *ci); |
de57606c SW |
712 | |
713 | /* what the mds thinks we want */ | |
c1944fed | 714 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
de57606c | 715 | |
37151668 YS |
716 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
717 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); | |
fe33032d YZ |
718 | extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, |
719 | struct ceph_mount_options *fsopt); | |
e30ee581 | 720 | extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 721 | struct ceph_cap_reservation *ctx, int need); |
7bf8f736 | 722 | extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 723 | struct ceph_cap_reservation *ctx); |
3d14c5d2 | 724 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
de57606c | 725 | int *total, int *avail, int *used, |
85ccce43 | 726 | int *reserved, int *min); |
de57606c | 727 | |
de57606c | 728 | |
de57606c SW |
729 | |
730 | /* | |
731 | * we keep buffered readdir results attached to file->private_data | |
732 | */ | |
4918b6d1 | 733 | #define CEPH_F_SYNC 1 |
9cfa1098 | 734 | #define CEPH_F_ATEND 2 |
4918b6d1 | 735 | |
de57606c | 736 | struct ceph_file_info { |
252c6728 SW |
737 | short fmode; /* initialized on open */ |
738 | short flags; /* CEPH_F_* */ | |
de57606c | 739 | |
5d988308 YZ |
740 | spinlock_t rw_contexts_lock; |
741 | struct list_head rw_contexts; | |
f4b97866 YZ |
742 | |
743 | errseq_t meta_err; | |
81f148a9 | 744 | u32 filp_gen; |
ff5d913d | 745 | atomic_t num_locks; |
bb48bd4d CX |
746 | }; |
747 | ||
748 | struct ceph_dir_file_info { | |
749 | struct ceph_file_info file_info; | |
5d988308 | 750 | |
de57606c SW |
751 | /* readdir: position within the dir */ |
752 | u32 frag; | |
753 | struct ceph_mds_request *last_readdir; | |
de57606c SW |
754 | |
755 | /* readdir: position within a frag */ | |
f0494206 | 756 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
de57606c | 757 | char *last_name; /* last entry in previous chunk */ |
fdd4e158 YZ |
758 | long long dir_release_count; |
759 | long long dir_ordered_count; | |
760 | int readdir_cache_idx; | |
de57606c SW |
761 | |
762 | /* used for -o dirstat read() on directory thing */ | |
763 | char *dir_info; | |
764 | int dir_info_len; | |
765 | }; | |
766 | ||
5d988308 YZ |
767 | struct ceph_rw_context { |
768 | struct list_head list; | |
769 | struct task_struct *thread; | |
770 | int caps; | |
771 | }; | |
772 | ||
773 | #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ | |
774 | struct ceph_rw_context _name = { \ | |
775 | .thread = current, \ | |
776 | .caps = _caps, \ | |
777 | } | |
778 | ||
779 | static inline void ceph_add_rw_context(struct ceph_file_info *cf, | |
780 | struct ceph_rw_context *ctx) | |
781 | { | |
782 | spin_lock(&cf->rw_contexts_lock); | |
783 | list_add(&ctx->list, &cf->rw_contexts); | |
784 | spin_unlock(&cf->rw_contexts_lock); | |
785 | } | |
786 | ||
787 | static inline void ceph_del_rw_context(struct ceph_file_info *cf, | |
788 | struct ceph_rw_context *ctx) | |
789 | { | |
790 | spin_lock(&cf->rw_contexts_lock); | |
791 | list_del(&ctx->list); | |
792 | spin_unlock(&cf->rw_contexts_lock); | |
793 | } | |
794 | ||
795 | static inline struct ceph_rw_context* | |
796 | ceph_find_rw_context(struct ceph_file_info *cf) | |
797 | { | |
798 | struct ceph_rw_context *ctx, *found = NULL; | |
799 | spin_lock(&cf->rw_contexts_lock); | |
800 | list_for_each_entry(ctx, &cf->rw_contexts, list) { | |
801 | if (ctx->thread == current) { | |
802 | found = ctx; | |
803 | break; | |
804 | } | |
805 | } | |
806 | spin_unlock(&cf->rw_contexts_lock); | |
807 | return found; | |
808 | } | |
809 | ||
fdd4e158 YZ |
810 | struct ceph_readdir_cache_control { |
811 | struct page *page; | |
812 | struct dentry **dentries; | |
813 | int index; | |
814 | }; | |
de57606c | 815 | |
de57606c SW |
816 | /* |
817 | * A "snap realm" describes a subset of the file hierarchy sharing | |
818 | * the same set of snapshots that apply to it. The realms themselves | |
819 | * are organized into a hierarchy, such that children inherit (some of) | |
820 | * the snapshots of their parents. | |
821 | * | |
822 | * All inodes within the realm that have capabilities are linked into a | |
823 | * per-realm list. | |
824 | */ | |
825 | struct ceph_snap_realm { | |
826 | u64 ino; | |
e3161f17 | 827 | struct inode *inode; |
de57606c | 828 | atomic_t nref; |
a105f00c SW |
829 | struct rb_node node; |
830 | ||
de57606c SW |
831 | u64 created, seq; |
832 | u64 parent_ino; | |
833 | u64 parent_since; /* snapid when our current parent became so */ | |
834 | ||
835 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ | |
aa711ee3 | 836 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
de57606c | 837 | u64 *snaps; /* snaps specific to this realm */ |
aa711ee3 | 838 | u32 num_snaps; |
de57606c SW |
839 | |
840 | struct ceph_snap_realm *parent; | |
841 | struct list_head children; /* list of child realms */ | |
842 | struct list_head child_item; | |
843 | ||
844 | struct list_head empty_item; /* if i have ref==0 */ | |
845 | ||
ae00d4f3 SW |
846 | struct list_head dirty_item; /* if realm needs new context */ |
847 | ||
de57606c SW |
848 | /* the current set of snaps for this realm */ |
849 | struct ceph_snap_context *cached_context; | |
850 | ||
851 | struct list_head inodes_with_caps; | |
852 | spinlock_t inodes_with_caps_lock; | |
853 | }; | |
854 | ||
3d14c5d2 | 855 | static inline int default_congestion_kb(void) |
de57606c | 856 | { |
3d14c5d2 YS |
857 | int congestion_kb; |
858 | ||
859 | /* | |
860 | * Copied from NFS | |
861 | * | |
862 | * congestion size, scale with available memory. | |
863 | * | |
864 | * 64MB: 8192k | |
865 | * 128MB: 11585k | |
866 | * 256MB: 16384k | |
867 | * 512MB: 23170k | |
868 | * 1GB: 32768k | |
869 | * 2GB: 46340k | |
870 | * 4GB: 65536k | |
871 | * 8GB: 92681k | |
872 | * 16GB: 131072k | |
873 | * | |
874 | * This allows larger machines to have larger/more transfers. | |
875 | * Limit the default to 256M | |
876 | */ | |
ca79b0c2 | 877 | congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); |
3d14c5d2 YS |
878 | if (congestion_kb > 256*1024) |
879 | congestion_kb = 256*1024; | |
880 | ||
881 | return congestion_kb; | |
de57606c SW |
882 | } |
883 | ||
884 | ||
d468e729 YZ |
885 | /* super.c */ |
886 | extern int ceph_force_reconnect(struct super_block *sb); | |
de57606c SW |
887 | /* snap.c */ |
888 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, | |
889 | u64 ino); | |
890 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, | |
891 | struct ceph_snap_realm *realm); | |
892 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, | |
893 | struct ceph_snap_realm *realm); | |
894 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, | |
982d6011 YZ |
895 | void *p, void *e, bool deletion, |
896 | struct ceph_snap_realm **realm_ret); | |
de57606c | 897 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
2600d2dd | 898 | struct ceph_mds_session *session, |
de57606c | 899 | struct ceph_msg *msg); |
fc837c8f | 900 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
de57606c SW |
901 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
902 | struct ceph_cap_snap *capsnap); | |
903 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); | |
904 | ||
75c9627e YZ |
905 | extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, |
906 | u64 snap); | |
907 | extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, | |
908 | struct ceph_snapid_map *sm); | |
909 | extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); | |
910 | extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); | |
911 | ||
912 | ||
de57606c SW |
913 | /* |
914 | * a cap_snap is "pending" if it is still awaiting an in-progress | |
915 | * sync write (that may/may not still update size, mtime, etc.). | |
916 | */ | |
917 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) | |
918 | { | |
919 | return !list_empty(&ci->i_cap_snaps) && | |
86056090 YZ |
920 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
921 | ci_item)->writing; | |
de57606c SW |
922 | } |
923 | ||
de57606c | 924 | /* inode.c */ |
966c7160 JL |
925 | struct ceph_mds_reply_info_in; |
926 | struct ceph_mds_reply_dirfrag; | |
927 | ||
de57606c SW |
928 | extern const struct inode_operations ceph_file_iops; |
929 | ||
930 | extern struct inode *ceph_alloc_inode(struct super_block *sb); | |
87bc5b89 | 931 | extern void ceph_evict_inode(struct inode *inode); |
cfa6d412 | 932 | extern void ceph_free_inode(struct inode *inode); |
de57606c SW |
933 | |
934 | extern struct inode *ceph_get_inode(struct super_block *sb, | |
935 | struct ceph_vino vino); | |
936 | extern struct inode *ceph_get_snapdir(struct inode *parent); | |
937 | extern int ceph_fill_file_size(struct inode *inode, int issued, | |
938 | u32 truncate_seq, u64 truncate_size, u64 size); | |
939 | extern void ceph_fill_file_time(struct inode *inode, int issued, | |
9bbeab41 AB |
940 | u64 time_warp_seq, struct timespec64 *ctime, |
941 | struct timespec64 *mtime, | |
942 | struct timespec64 *atime); | |
966c7160 JL |
943 | extern int ceph_fill_inode(struct inode *inode, struct page *locked_page, |
944 | struct ceph_mds_reply_info_in *iinfo, | |
945 | struct ceph_mds_reply_dirfrag *dirinfo, | |
946 | struct ceph_mds_session *session, int cap_fmode, | |
947 | struct ceph_cap_reservation *caps_reservation); | |
de57606c | 948 | extern int ceph_fill_trace(struct super_block *sb, |
f5a03b08 | 949 | struct ceph_mds_request *req); |
de57606c SW |
950 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
951 | struct ceph_mds_session *session); | |
952 | ||
953 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); | |
954 | ||
efb0ca76 | 955 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
b415bf4f | 956 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
3c6f6b79 | 957 | extern void ceph_queue_vmtruncate(struct inode *inode); |
3c6f6b79 SW |
958 | extern void ceph_queue_invalidate(struct inode *inode); |
959 | extern void ceph_queue_writeback(struct inode *inode); | |
3e1d0452 | 960 | extern void ceph_async_iput(struct inode *inode); |
de57606c | 961 | |
01deead0 YZ |
962 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
963 | int mask, bool force); | |
964 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) | |
965 | { | |
966 | return __ceph_do_getattr(inode, NULL, mask, force); | |
967 | } | |
10556cb2 | 968 | extern int ceph_permission(struct inode *inode, int mask); |
a26fecca | 969 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
de57606c | 970 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
a528d35e DH |
971 | extern int ceph_getattr(const struct path *path, struct kstat *stat, |
972 | u32 request_mask, unsigned int flags); | |
de57606c SW |
973 | |
974 | /* xattr.c */ | |
a26fecca | 975 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
7221fe4c | 976 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
de57606c | 977 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
12fe3dda | 978 | extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
de57606c | 979 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); |
b1ee94aa | 980 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
de57606c | 981 | |
5c31e92d YZ |
982 | struct ceph_acl_sec_ctx { |
983 | #ifdef CONFIG_CEPH_FS_POSIX_ACL | |
984 | void *default_acl; | |
985 | void *acl; | |
ac6713cc YZ |
986 | #endif |
987 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL | |
988 | void *sec_ctx; | |
989 | u32 sec_ctxlen; | |
5c31e92d YZ |
990 | #endif |
991 | struct ceph_pagelist *pagelist; | |
992 | }; | |
993 | ||
315f2408 YZ |
994 | #ifdef CONFIG_SECURITY |
995 | extern bool ceph_security_xattr_deadlock(struct inode *in); | |
996 | extern bool ceph_security_xattr_wanted(struct inode *in); | |
997 | #else | |
998 | static inline bool ceph_security_xattr_deadlock(struct inode *in) | |
999 | { | |
1000 | return false; | |
1001 | } | |
1002 | static inline bool ceph_security_xattr_wanted(struct inode *in) | |
1003 | { | |
1004 | return false; | |
1005 | } | |
1006 | #endif | |
1007 | ||
ac6713cc YZ |
1008 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL |
1009 | extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
1010 | struct ceph_acl_sec_ctx *ctx); | |
668959a5 JL |
1011 | static inline void ceph_security_invalidate_secctx(struct inode *inode) |
1012 | { | |
1013 | security_inode_invalidate_secctx(inode); | |
1014 | } | |
ac6713cc YZ |
1015 | #else |
1016 | static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
1017 | struct ceph_acl_sec_ctx *ctx) | |
1018 | { | |
1019 | return 0; | |
1020 | } | |
1021 | static inline void ceph_security_invalidate_secctx(struct inode *inode) | |
1022 | { | |
1023 | } | |
1024 | #endif | |
1025 | ||
5c31e92d | 1026 | void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); |
7221fe4c | 1027 | |
5c31e92d | 1028 | /* acl.c */ |
7221fe4c GZ |
1029 | #ifdef CONFIG_CEPH_FS_POSIX_ACL |
1030 | ||
1031 | struct posix_acl *ceph_get_acl(struct inode *, int); | |
72466d0b | 1032 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
b1ee94aa | 1033 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d YZ |
1034 | struct ceph_acl_sec_ctx *as_ctx); |
1035 | void ceph_init_inode_acls(struct inode *inode, | |
1036 | struct ceph_acl_sec_ctx *as_ctx); | |
c969d9bf GZ |
1037 | |
1038 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
1039 | { | |
1040 | forget_all_cached_acls(inode); | |
1041 | } | |
7221fe4c GZ |
1042 | |
1043 | #else | |
1044 | ||
1045 | #define ceph_get_acl NULL | |
72466d0b | 1046 | #define ceph_set_acl NULL |
7221fe4c | 1047 | |
b1ee94aa | 1048 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d | 1049 | struct ceph_acl_sec_ctx *as_ctx) |
7221fe4c GZ |
1050 | { |
1051 | return 0; | |
1052 | } | |
b1ee94aa | 1053 | static inline void ceph_init_inode_acls(struct inode *inode, |
5c31e92d | 1054 | struct ceph_acl_sec_ctx *as_ctx) |
b1ee94aa YZ |
1055 | { |
1056 | } | |
7221fe4c GZ |
1057 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
1058 | { | |
1059 | return 0; | |
1060 | } | |
1061 | ||
1062 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
1063 | { | |
1064 | } | |
1065 | ||
1066 | #endif | |
1067 | ||
de57606c SW |
1068 | /* caps.c */ |
1069 | extern const char *ceph_cap_string(int c); | |
1070 | extern void ceph_handle_caps(struct ceph_mds_session *session, | |
1071 | struct ceph_msg *msg); | |
d9df2783 YZ |
1072 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
1073 | struct ceph_cap_reservation *ctx); | |
1074 | extern void ceph_add_cap(struct inode *inode, | |
1075 | struct ceph_mds_session *session, u64 cap_id, | |
135e671e | 1076 | unsigned issued, unsigned wanted, |
d9df2783 YZ |
1077 | unsigned cap, unsigned seq, u64 realmino, int flags, |
1078 | struct ceph_cap **new_cap); | |
a096b09a | 1079 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
d6e47819 | 1080 | extern void __ceph_remove_caps(struct ceph_inode_info *ci); |
37151668 YS |
1081 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
1082 | struct ceph_cap *cap); | |
9215aeea | 1083 | extern int ceph_is_any_caps(struct inode *inode); |
de57606c | 1084 | |
f1a3d572 | 1085 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
02c24a82 JB |
1086 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
1087 | int datasync); | |
e548e9b9 YZ |
1088 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1089 | struct ceph_mds_session *session); | |
de57606c SW |
1090 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1091 | struct ceph_mds_session *session); | |
e8a4d267 JL |
1092 | void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, |
1093 | struct ceph_inode_info *ci); | |
2bc50259 GF |
1094 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
1095 | int mds); | |
40dcf75e JL |
1096 | extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, |
1097 | bool snap_rwsem_locked); | |
de57606c SW |
1098 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); |
1099 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); | |
e64f44a8 XL |
1100 | extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, |
1101 | int had); | |
de57606c SW |
1102 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, |
1103 | struct ceph_snap_context *snapc); | |
ed9b430c YZ |
1104 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
1105 | struct ceph_mds_session **psession); | |
efb0ca76 | 1106 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
de57606c SW |
1107 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
1108 | struct ceph_mds_session *session); | |
afcdaea3 SW |
1109 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
1110 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); | |
6ef0bc6d | 1111 | extern int ceph_drop_caps_for_unlink(struct inode *inode); |
de57606c SW |
1112 | extern int ceph_encode_inode_release(void **p, struct inode *inode, |
1113 | int mds, int drop, int unless, int force); | |
1114 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, | |
ca6c8ae0 | 1115 | struct inode *dir, |
de57606c SW |
1116 | int mds, int drop, int unless); |
1117 | ||
5e3ded1b | 1118 | extern int ceph_get_caps(struct file *filp, int need, int want, |
31c542a1 | 1119 | loff_t endoff, int *got, struct page **pinned_page); |
5e3ded1b | 1120 | extern int ceph_try_get_caps(struct inode *inode, |
2ee9dd95 | 1121 | int need, int want, bool nonblock, int *got); |
de57606c SW |
1122 | |
1123 | /* for counting open files by mode */ | |
719a2514 YZ |
1124 | extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count); |
1125 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count); | |
1126 | extern void __ceph_touch_fmode(struct ceph_inode_info *ci, | |
1127 | struct ceph_mds_client *mdsc, int fmode); | |
de57606c SW |
1128 | |
1129 | /* addr.c */ | |
1130 | extern const struct address_space_operations ceph_aops; | |
1131 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); | |
10183a69 | 1132 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
5e3ded1b | 1133 | extern int ceph_pool_perm_check(struct inode *inode, int need); |
10183a69 | 1134 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); |
de57606c SW |
1135 | |
1136 | /* file.c */ | |
1137 | extern const struct file_operations ceph_file_fops; | |
9e0eb85d | 1138 | |
719a2514 | 1139 | extern int ceph_renew_caps(struct inode *inode, int fmode); |
de57606c | 1140 | extern int ceph_open(struct inode *inode, struct file *file); |
5ef50c3b | 1141 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
44907d79 | 1142 | struct file *file, unsigned flags, umode_t mode); |
de57606c | 1143 | extern int ceph_release(struct inode *inode, struct file *filp); |
31c542a1 YZ |
1144 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
1145 | char *data, size_t len); | |
55f2a045 | 1146 | |
de57606c SW |
1147 | /* dir.c */ |
1148 | extern const struct file_operations ceph_dir_fops; | |
38c48b5f | 1149 | extern const struct file_operations ceph_snapdir_fops; |
de57606c | 1150 | extern const struct inode_operations ceph_dir_iops; |
38c48b5f | 1151 | extern const struct inode_operations ceph_snapdir_iops; |
18fc8abd | 1152 | extern const struct dentry_operations ceph_dentry_ops; |
de57606c | 1153 | |
f3c4ebe6 | 1154 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
de57606c | 1155 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
468640e3 SW |
1156 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
1157 | struct dentry *dentry, int err); | |
de57606c SW |
1158 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
1159 | struct dentry *dentry, int err); | |
1160 | ||
37c4efc1 YZ |
1161 | extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); |
1162 | extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); | |
81a6cf2d | 1163 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
37c4efc1 | 1164 | extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); |
e5f86dc3 | 1165 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
fdd4e158 | 1166 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
de57606c | 1167 | |
de57606c SW |
1168 | /* ioctl.c */ |
1169 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); | |
1170 | ||
1171 | /* export.c */ | |
1172 | extern const struct export_operations ceph_export_ops; | |
3886274a | 1173 | struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); |
de57606c | 1174 | |
40819f6f | 1175 | /* locks.c */ |
eb13e832 | 1176 | extern __init void ceph_flock_init(void); |
40819f6f GF |
1177 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
1178 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); | |
1179 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); | |
39be95e9 JS |
1180 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
1181 | struct ceph_filelock *flocks, | |
1182 | int num_fcntl_locks, | |
1183 | int num_flock_locks); | |
1184 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, | |
1185 | struct ceph_pagelist *pagelist, | |
1186 | int num_fcntl_locks, int num_flock_locks); | |
40819f6f | 1187 | |
3d14c5d2 | 1188 | /* debugfs.c */ |
1a829ff2 | 1189 | extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); |
3d14c5d2 YS |
1190 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); |
1191 | ||
fb18a575 | 1192 | /* quota.c */ |
d557c48d LH |
1193 | static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) |
1194 | { | |
1195 | return ci->i_max_files || ci->i_max_bytes; | |
1196 | } | |
1197 | ||
1198 | extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); | |
1199 | ||
1200 | static inline void __ceph_update_quota(struct ceph_inode_info *ci, | |
1201 | u64 max_bytes, u64 max_files) | |
1202 | { | |
1203 | bool had_quota, has_quota; | |
1204 | had_quota = __ceph_has_any_quota(ci); | |
1205 | ci->i_max_bytes = max_bytes; | |
1206 | ci->i_max_files = max_files; | |
1207 | has_quota = __ceph_has_any_quota(ci); | |
1208 | ||
1209 | if (had_quota != has_quota) | |
1210 | ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); | |
1211 | } | |
1212 | ||
fb18a575 LH |
1213 | extern void ceph_handle_quota(struct ceph_mds_client *mdsc, |
1214 | struct ceph_mds_session *session, | |
1215 | struct ceph_msg *msg); | |
b7a29217 | 1216 | extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); |
2b83845f LH |
1217 | extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, |
1218 | loff_t newlen); | |
1ab302a0 LH |
1219 | extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, |
1220 | loff_t newlen); | |
9122eed5 LH |
1221 | extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, |
1222 | struct kstatfs *buf); | |
dffdcd71 LH |
1223 | extern int ceph_quota_check_rename(struct ceph_mds_client *mdsc, |
1224 | struct inode *old, struct inode *new); | |
0c44a8e0 | 1225 | extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); |
fb18a575 | 1226 | |
de57606c | 1227 | #endif /* _FS_CEPH_SUPER_H */ |