]>
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 | 120 | |
e3ec8d68 YZ |
121 | atomic_long_t writeback_count; |
122 | ||
1cf89a8d | 123 | struct workqueue_struct *inode_wq; |
e3ec8d68 | 124 | struct workqueue_struct *cap_wq; |
de57606c | 125 | |
0743304d | 126 | #ifdef CONFIG_DEBUG_FS |
3d14c5d2 | 127 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
2baba250 | 128 | struct dentry *debugfs_congestion_kb; |
06edf046 | 129 | struct dentry *debugfs_bdi; |
3d14c5d2 | 130 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
f9009efa | 131 | struct dentry *debugfs_metric; |
14ed9703 | 132 | struct dentry *debugfs_mds_sessions; |
0743304d | 133 | #endif |
99ccbd22 MT |
134 | |
135 | #ifdef CONFIG_CEPH_FSCACHE | |
136 | struct fscache_cookie *fscache; | |
99ccbd22 | 137 | #endif |
de57606c SW |
138 | }; |
139 | ||
3d14c5d2 | 140 | |
de57606c SW |
141 | /* |
142 | * File i/o capability. This tracks shared state with the metadata | |
143 | * server that allows us to cache or writeback attributes or to read | |
144 | * and write data. For any given inode, we should have one or more | |
145 | * capabilities, one issued by each metadata server, and our | |
146 | * cumulative access is the OR of all issued capabilities. | |
147 | * | |
148 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds | |
149 | * session capability lists. | |
150 | */ | |
151 | struct ceph_cap { | |
152 | struct ceph_inode_info *ci; | |
153 | struct rb_node ci_node; /* per-ci cap tree */ | |
154 | struct ceph_mds_session *session; | |
155 | struct list_head session_caps; /* per-session caplist */ | |
de57606c | 156 | u64 cap_id; /* unique cap id (mds provided) */ |
745a8e3b YZ |
157 | union { |
158 | /* in-use caps */ | |
159 | struct { | |
160 | int issued; /* latest, from the mds */ | |
161 | int implemented; /* implemented superset of | |
162 | issued (for revocation) */ | |
163 | int mds, mds_wanted; | |
164 | }; | |
165 | /* caps to release */ | |
166 | struct { | |
167 | u64 cap_ino; | |
168 | int queue_release; | |
169 | }; | |
170 | }; | |
685f9a5d SW |
171 | u32 seq, issue_seq, mseq; |
172 | u32 cap_gen; /* active/stale cycle */ | |
de57606c SW |
173 | unsigned long last_used; |
174 | struct list_head caps_item; | |
175 | }; | |
176 | ||
a0d93e32 YZ |
177 | #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */ |
178 | #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */ | |
179 | #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */ | |
de57606c | 180 | |
0e294387 YZ |
181 | struct ceph_cap_flush { |
182 | u64 tid; | |
183 | int caps; /* 0 means capsnap */ | |
c8799fc4 | 184 | bool wake; /* wake up flush waiters when finish ? */ |
0e294387 YZ |
185 | struct list_head g_list; // global |
186 | struct list_head i_list; // per inode | |
187 | }; | |
188 | ||
de57606c SW |
189 | /* |
190 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, | |
191 | * we first complete any in-process sync writes and writeback any dirty | |
192 | * data before flushing the snapped state (tracked here) back to the MDS. | |
193 | */ | |
194 | struct ceph_cap_snap { | |
805692d0 | 195 | refcount_t nref; |
0e294387 | 196 | struct list_head ci_item; |
de57606c | 197 | |
0e294387 YZ |
198 | struct ceph_cap_flush cap_flush; |
199 | ||
200 | u64 follows; | |
de57606c SW |
201 | int issued, dirty; |
202 | struct ceph_snap_context *context; | |
203 | ||
5706b27d | 204 | umode_t mode; |
05cb11c1 EB |
205 | kuid_t uid; |
206 | kgid_t gid; | |
de57606c | 207 | |
4a625be4 | 208 | struct ceph_buffer *xattr_blob; |
de57606c SW |
209 | u64 xattr_version; |
210 | ||
211 | u64 size; | |
176c77c9 | 212 | u64 change_attr; |
ec62b894 | 213 | struct timespec64 mtime, atime, ctime, btime; |
de57606c | 214 | u64 time_warp_seq; |
5f743e45 YZ |
215 | u64 truncate_size; |
216 | u32 truncate_seq; | |
de57606c SW |
217 | int writing; /* a sync write is still in progress */ |
218 | int dirty_pages; /* dirty pages awaiting writeback */ | |
e20d258d | 219 | bool inline_data; |
86056090 | 220 | bool need_flush; |
de57606c SW |
221 | }; |
222 | ||
223 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) | |
224 | { | |
805692d0 | 225 | if (refcount_dec_and_test(&capsnap->nref)) { |
4a625be4 SW |
226 | if (capsnap->xattr_blob) |
227 | ceph_buffer_put(capsnap->xattr_blob); | |
de57606c | 228 | kfree(capsnap); |
4a625be4 | 229 | } |
de57606c SW |
230 | } |
231 | ||
232 | /* | |
233 | * The frag tree describes how a directory is fragmented, potentially across | |
234 | * multiple metadata servers. It is also used to indicate points where | |
235 | * metadata authority is delegated, and whether/where metadata is replicated. | |
236 | * | |
237 | * A _leaf_ frag will be present in the i_fragtree IFF there is | |
238 | * delegation info. That is, if mds >= 0 || ndist > 0. | |
239 | */ | |
240 | #define CEPH_MAX_DIRFRAG_REP 4 | |
241 | ||
242 | struct ceph_inode_frag { | |
243 | struct rb_node node; | |
244 | ||
245 | /* fragtree state */ | |
246 | u32 frag; | |
247 | int split_by; /* i.e. 2^(split_by) children */ | |
248 | ||
249 | /* delegation and replication info */ | |
250 | int mds; /* -1 if same authority as parent */ | |
251 | int ndist; /* >0 if replicated */ | |
252 | int dist[CEPH_MAX_DIRFRAG_REP]; | |
253 | }; | |
254 | ||
255 | /* | |
256 | * We cache inode xattrs as an encoded blob until they are first used, | |
257 | * at which point we parse them into an rbtree. | |
258 | */ | |
259 | struct ceph_inode_xattr { | |
260 | struct rb_node node; | |
261 | ||
262 | const char *name; | |
263 | int name_len; | |
264 | const char *val; | |
265 | int val_len; | |
266 | int dirty; | |
267 | ||
268 | int should_free_name; | |
269 | int should_free_val; | |
270 | }; | |
271 | ||
3d14c5d2 YS |
272 | /* |
273 | * Ceph dentry state | |
274 | */ | |
275 | struct ceph_dentry_info { | |
37c4efc1 | 276 | struct dentry *dentry; |
3d14c5d2 | 277 | struct ceph_mds_session *lease_session; |
37c4efc1 YZ |
278 | struct list_head lease_list; |
279 | unsigned flags; | |
97aeb6bf YZ |
280 | int lease_shared_gen; |
281 | u32 lease_gen; | |
3d14c5d2 YS |
282 | u32 lease_seq; |
283 | unsigned long lease_renew_after, lease_renew_from; | |
9b16f03c | 284 | unsigned long time; |
3d14c5d2 YS |
285 | u64 offset; |
286 | }; | |
287 | ||
37c4efc1 YZ |
288 | #define CEPH_DENTRY_REFERENCED 1 |
289 | #define CEPH_DENTRY_LEASE_LIST 2 | |
290 | #define CEPH_DENTRY_SHRINK_LIST 4 | |
f5e17aed | 291 | #define CEPH_DENTRY_PRIMARY_LINK 8 |
37c4efc1 | 292 | |
de57606c SW |
293 | struct ceph_inode_xattrs_info { |
294 | /* | |
295 | * (still encoded) xattr blob. we avoid the overhead of parsing | |
296 | * this until someone actually calls getxattr, etc. | |
297 | * | |
298 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == | |
299 | * NULL means we don't know. | |
300 | */ | |
301 | struct ceph_buffer *blob, *prealloc_blob; | |
302 | ||
303 | struct rb_root index; | |
304 | bool dirty; | |
305 | int count; | |
306 | int names_size; | |
307 | int vals_size; | |
308 | u64 version, index_version; | |
309 | }; | |
310 | ||
311 | /* | |
312 | * Ceph inode. | |
313 | */ | |
de57606c SW |
314 | struct ceph_inode_info { |
315 | struct ceph_vino i_vino; /* ceph ino + snap */ | |
316 | ||
be655596 SW |
317 | spinlock_t i_ceph_lock; |
318 | ||
de57606c | 319 | u64 i_version; |
31c542a1 | 320 | u64 i_inline_version; |
de57606c SW |
321 | u32 i_time_warp_seq; |
322 | ||
891f3f5a | 323 | unsigned long i_ceph_flags; |
fdd4e158 YZ |
324 | atomic64_t i_release_count; |
325 | atomic64_t i_ordered_count; | |
326 | atomic64_t i_complete_seq[2]; | |
de57606c | 327 | |
6c0f3af7 | 328 | struct ceph_dir_layout i_dir_layout; |
de57606c | 329 | struct ceph_file_layout i_layout; |
785892fe | 330 | struct ceph_file_layout i_cached_layout; // for async creates |
de57606c SW |
331 | char *i_symlink; |
332 | ||
333 | /* for dirs */ | |
9bbeab41 | 334 | struct timespec64 i_rctime; |
de57606c SW |
335 | u64 i_rbytes, i_rfiles, i_rsubdirs; |
336 | u64 i_files, i_subdirs; | |
de57606c | 337 | |
fb18a575 LH |
338 | /* quotas */ |
339 | u64 i_max_bytes, i_max_files; | |
340 | ||
08796873 YZ |
341 | s32 i_dir_pin; |
342 | ||
de57606c | 343 | struct rb_root i_fragtree; |
1b1bc16d | 344 | int i_fragtree_nsplits; |
de57606c SW |
345 | struct mutex i_fragtree_mutex; |
346 | ||
347 | struct ceph_inode_xattrs_info i_xattrs; | |
348 | ||
be655596 | 349 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
de57606c SW |
350 | * s_mutex. */ |
351 | struct rb_root i_caps; /* cap list */ | |
352 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ | |
353 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ | |
1cf03a68 JL |
354 | |
355 | /* | |
f1f565a2 | 356 | * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty |
1cf03a68 JL |
357 | * is protected by the mdsc->cap_dirty_lock, but each individual item |
358 | * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty | |
359 | * requires the mdsc->cap_dirty_lock. List presence for an item can | |
360 | * be tested under the i_ceph_lock. Changing anything requires both. | |
361 | */ | |
362 | struct list_head i_dirty_item; | |
363 | ||
364 | /* | |
829ad4db JL |
365 | * Link to session's s_cap_flushing list. Protected in a similar |
366 | * fashion to i_dirty_item, but also by the s_mutex for changes. The | |
367 | * s_cap_flushing list can be walked while holding either the s_mutex | |
368 | * or msdc->cap_dirty_lock. List presence can also be checked while | |
369 | * holding the i_ceph_lock for this inode. | |
1cf03a68 JL |
370 | */ |
371 | struct list_head i_flushing_item; | |
372 | ||
de57606c SW |
373 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
374 | * overlapping, pipelined cap flushes to the mds. we can probably | |
375 | * reduce the tid to 8 bits if we're concerned about inode size. */ | |
f66fd9f0 | 376 | struct ceph_cap_flush *i_prealloc_cap_flush; |
e4500b5e | 377 | struct list_head i_cap_flush_list; |
de57606c | 378 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
de57606c SW |
379 | unsigned long i_hold_caps_max; /* jiffies */ |
380 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ | |
de57606c SW |
381 | struct ceph_cap_reservation i_cap_migration_resv; |
382 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ | |
7d8cb26d SW |
383 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
384 | dirty|flushing caps */ | |
de57606c SW |
385 | unsigned i_snap_caps; /* cap bits for snapped files */ |
386 | ||
719a2514 YZ |
387 | unsigned long i_last_rd; |
388 | unsigned long i_last_wr; | |
774a6a11 | 389 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
de57606c | 390 | |
b0d7c223 | 391 | struct mutex i_truncate_mutex; |
de57606c SW |
392 | u32 i_truncate_seq; /* last truncate to smaller size */ |
393 | u64 i_truncate_size; /* and the size we last truncated down to */ | |
394 | int i_truncate_pending; /* still need to call vmtruncate */ | |
395 | ||
396 | u64 i_max_size; /* max file size authorized by mds */ | |
397 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ | |
398 | u64 i_wanted_max_size; /* offset we'd like to write too */ | |
399 | u64 i_requested_max_size; /* max_size we've requested */ | |
400 | ||
401 | /* held references to caps */ | |
402 | int i_pin_ref; | |
f85122af | 403 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref; |
de57606c | 404 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
89aa5930 | 405 | atomic_t i_filelock_ref; |
97aeb6bf | 406 | atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ |
cd045cb4 | 407 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
de57606c SW |
408 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
409 | ||
de57606c | 410 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
68cd5b4b | 411 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
de57606c SW |
412 | spinlock_t i_unsafe_lock; |
413 | ||
75c9627e YZ |
414 | union { |
415 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ | |
416 | struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ | |
417 | }; | |
de57606c SW |
418 | int i_snap_realm_counter; /* snap realm (if caps) */ |
419 | struct list_head i_snap_realm_item; | |
420 | struct list_head i_snap_flush_item; | |
245ce991 | 421 | struct timespec64 i_btime; |
193e7b37 | 422 | struct timespec64 i_snap_btime; |
de57606c | 423 | |
1cf89a8d YZ |
424 | struct work_struct i_work; |
425 | unsigned long i_work_mask; | |
de57606c | 426 | |
99ccbd22 MT |
427 | #ifdef CONFIG_CEPH_FSCACHE |
428 | struct fscache_cookie *fscache; | |
f7f7e7a0 | 429 | u32 i_fscache_gen; |
99ccbd22 | 430 | #endif |
f4b97866 YZ |
431 | errseq_t i_meta_err; |
432 | ||
de57606c SW |
433 | struct inode vfs_inode; /* at end */ |
434 | }; | |
435 | ||
721d5c13 JL |
436 | static inline struct ceph_inode_info * |
437 | ceph_inode(const struct inode *inode) | |
de57606c | 438 | { |
fbbccec9 | 439 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
de57606c SW |
440 | } |
441 | ||
721d5c13 JL |
442 | static inline struct ceph_fs_client * |
443 | ceph_inode_to_client(const struct inode *inode) | |
ad1fee96 YS |
444 | { |
445 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; | |
446 | } | |
447 | ||
721d5c13 JL |
448 | static inline struct ceph_fs_client * |
449 | ceph_sb_to_client(const struct super_block *sb) | |
ad1fee96 YS |
450 | { |
451 | return (struct ceph_fs_client *)sb->s_fs_info; | |
452 | } | |
453 | ||
721d5c13 JL |
454 | static inline struct ceph_vino |
455 | ceph_vino(const struct inode *inode) | |
3d14c5d2 YS |
456 | { |
457 | return ceph_inode(inode)->i_vino; | |
458 | } | |
459 | ||
ebce3eb2 | 460 | static inline u32 ceph_ino_to_ino32(u64 vino) |
ad1fee96 | 461 | { |
3310f754 AO |
462 | u32 ino = vino & 0xffffffff; |
463 | ino ^= vino >> 32; | |
ad1fee96 | 464 | if (!ino) |
a661fc56 | 465 | ino = 2; |
ad1fee96 YS |
466 | return ino; |
467 | } | |
468 | ||
469 | /* | |
ebce3eb2 JL |
470 | * Inode numbers in cephfs are 64 bits, but inode->i_ino is 32-bits on |
471 | * some arches. We generally do not use this value inside the ceph driver, but | |
472 | * we do want to set it to something, so that generic vfs code has an | |
473 | * appropriate value for tracepoints and the like. | |
3d14c5d2 | 474 | */ |
ebce3eb2 | 475 | static inline ino_t ceph_vino_to_ino_t(struct ceph_vino vino) |
3d14c5d2 | 476 | { |
ebce3eb2 JL |
477 | if (sizeof(ino_t) == sizeof(u32)) |
478 | return ceph_ino_to_ino32(vino.ino); | |
3310f754 | 479 | return (ino_t)vino.ino; |
ad1fee96 | 480 | } |
ad1fee96 | 481 | |
3d14c5d2 YS |
482 | /* for printf-style formatting */ |
483 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap | |
484 | ||
485 | static inline u64 ceph_ino(struct inode *inode) | |
486 | { | |
487 | return ceph_inode(inode)->i_vino.ino; | |
488 | } | |
ebce3eb2 | 489 | |
3d14c5d2 YS |
490 | static inline u64 ceph_snap(struct inode *inode) |
491 | { | |
492 | return ceph_inode(inode)->i_vino.snap; | |
493 | } | |
494 | ||
ebce3eb2 JL |
495 | /** |
496 | * ceph_present_ino - format an inode number for presentation to userland | |
497 | * @sb: superblock where the inode lives | |
498 | * @ino: inode number to (possibly) convert | |
499 | * | |
500 | * If the user mounted with the ino32 option, then the 64-bit value needs | |
501 | * to be converted to something that can fit inside 32 bits. Note that | |
502 | * internal kernel code never uses this value, so this is entirely for | |
503 | * userland consumption. | |
504 | */ | |
505 | static inline u64 ceph_present_ino(struct super_block *sb, u64 ino) | |
506 | { | |
507 | if (unlikely(ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))) | |
508 | return ceph_ino_to_ino32(ino); | |
509 | return ino; | |
510 | } | |
511 | ||
512 | static inline u64 ceph_present_inode(struct inode *inode) | |
513 | { | |
514 | return ceph_present_ino(inode->i_sb, ceph_ino(inode)); | |
515 | } | |
516 | ||
3d14c5d2 YS |
517 | static inline int ceph_ino_compare(struct inode *inode, void *data) |
518 | { | |
519 | struct ceph_vino *pvino = (struct ceph_vino *)data; | |
520 | struct ceph_inode_info *ci = ceph_inode(inode); | |
521 | return ci->i_vino.ino == pvino->ino && | |
522 | ci->i_vino.snap == pvino->snap; | |
523 | } | |
524 | ||
ebce3eb2 | 525 | |
3d14c5d2 YS |
526 | static inline struct inode *ceph_find_inode(struct super_block *sb, |
527 | struct ceph_vino vino) | |
528 | { | |
ebce3eb2 JL |
529 | /* |
530 | * NB: The hashval will be run through the fs/inode.c hash function | |
531 | * anyway, so there is no need to squash the inode number down to | |
532 | * 32-bits first. Just use low-order bits on arches with 32-bit long. | |
533 | */ | |
534 | return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino); | |
3d14c5d2 YS |
535 | } |
536 | ||
537 | ||
538 | /* | |
539 | * Ceph inode. | |
540 | */ | |
10183a69 | 541 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
10183a69 | 542 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ |
daca8bda JL |
543 | #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */ |
544 | #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */ | |
545 | #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */ | |
546 | #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */ | |
c0e385b1 YZ |
547 | #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */ |
548 | #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */ | |
549 | #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */ | |
550 | #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */ | |
551 | #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */ | |
552 | #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */ | |
891f3f5a | 553 | #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT) |
26544c62 | 554 | |
1cf89a8d YZ |
555 | /* |
556 | * Masks of ceph inode work. | |
557 | */ | |
558 | #define CEPH_I_WORK_WRITEBACK 0 /* writeback */ | |
559 | #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */ | |
560 | #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */ | |
561 | ||
26544c62 JL |
562 | /* |
563 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode | |
564 | * and then clear it when they start succeeding. Note that we do a lockless | |
565 | * check first, and only take the lock if it looks like it needs to be changed. | |
566 | * The write submission code just takes this as a hint, so we're not too | |
567 | * worried if a few slip through in either direction. | |
568 | */ | |
569 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) | |
570 | { | |
571 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { | |
572 | spin_lock(&ci->i_ceph_lock); | |
573 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; | |
574 | spin_unlock(&ci->i_ceph_lock); | |
575 | } | |
576 | } | |
577 | ||
578 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) | |
579 | { | |
580 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { | |
581 | spin_lock(&ci->i_ceph_lock); | |
582 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; | |
583 | spin_unlock(&ci->i_ceph_lock); | |
584 | } | |
585 | } | |
3d14c5d2 | 586 | |
2f276c51 | 587 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
fdd4e158 YZ |
588 | long long release_count, |
589 | long long ordered_count) | |
de57606c | 590 | { |
74960773 AP |
591 | /* |
592 | * Makes sure operations that setup readdir cache (update page | |
593 | * cache and i_size) are strongly ordered w.r.t. the following | |
594 | * atomic64_set() operations. | |
595 | */ | |
596 | smp_mb(); | |
fdd4e158 YZ |
597 | atomic64_set(&ci->i_complete_seq[0], release_count); |
598 | atomic64_set(&ci->i_complete_seq[1], ordered_count); | |
de57606c SW |
599 | } |
600 | ||
2f276c51 | 601 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
de57606c | 602 | { |
fdd4e158 YZ |
603 | atomic64_inc(&ci->i_release_count); |
604 | } | |
605 | ||
606 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) | |
607 | { | |
608 | atomic64_inc(&ci->i_ordered_count); | |
2f276c51 | 609 | } |
de57606c | 610 | |
2f276c51 YZ |
611 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
612 | { | |
fdd4e158 YZ |
613 | return atomic64_read(&ci->i_complete_seq[0]) == |
614 | atomic64_read(&ci->i_release_count); | |
de57606c SW |
615 | } |
616 | ||
70db4f36 YZ |
617 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
618 | { | |
fdd4e158 YZ |
619 | return atomic64_read(&ci->i_complete_seq[0]) == |
620 | atomic64_read(&ci->i_release_count) && | |
621 | atomic64_read(&ci->i_complete_seq[1]) == | |
622 | atomic64_read(&ci->i_ordered_count); | |
70db4f36 YZ |
623 | } |
624 | ||
2f276c51 | 625 | static inline void ceph_dir_clear_complete(struct inode *inode) |
de57606c | 626 | { |
2f276c51 YZ |
627 | __ceph_dir_clear_complete(ceph_inode(inode)); |
628 | } | |
de57606c | 629 | |
70db4f36 | 630 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
2f276c51 | 631 | { |
fdd4e158 | 632 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
de57606c SW |
633 | } |
634 | ||
70db4f36 YZ |
635 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
636 | { | |
fdd4e158 YZ |
637 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
638 | smp_rmb(); | |
70db4f36 YZ |
639 | return ret; |
640 | } | |
de57606c SW |
641 | |
642 | /* find a specific frag @f */ | |
643 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, | |
644 | u32 f); | |
645 | ||
646 | /* | |
647 | * choose fragment for value @v. copy frag content to pfrag, if leaf | |
648 | * exists | |
649 | */ | |
650 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, | |
651 | struct ceph_inode_frag *pfrag, | |
652 | int *found); | |
653 | ||
1e9c2eb6 | 654 | static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) |
de57606c SW |
655 | { |
656 | return (struct ceph_dentry_info *)dentry->d_fsdata; | |
657 | } | |
658 | ||
de57606c SW |
659 | /* |
660 | * caps helpers | |
661 | */ | |
662 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) | |
663 | { | |
664 | return !RB_EMPTY_ROOT(&ci->i_caps); | |
665 | } | |
666 | ||
667 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); | |
668 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); | |
1af16d54 XL |
669 | extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask, |
670 | int t); | |
de57606c SW |
671 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, |
672 | struct ceph_cap *cap); | |
673 | ||
674 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) | |
675 | { | |
676 | int issued; | |
be655596 | 677 | spin_lock(&ci->i_ceph_lock); |
de57606c | 678 | issued = __ceph_caps_issued(ci, NULL); |
be655596 | 679 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
680 | return issued; |
681 | } | |
682 | ||
1af16d54 XL |
683 | static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, |
684 | int mask, int touch) | |
de57606c SW |
685 | { |
686 | int r; | |
be655596 | 687 | spin_lock(&ci->i_ceph_lock); |
1af16d54 | 688 | r = __ceph_caps_issued_mask_metric(ci, mask, touch); |
be655596 | 689 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
690 | return r; |
691 | } | |
692 | ||
693 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) | |
694 | { | |
695 | return ci->i_dirty_caps | ci->i_flushing_caps; | |
696 | } | |
f66fd9f0 YZ |
697 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
698 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); | |
699 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, | |
700 | struct ceph_cap_flush **pcf); | |
de57606c | 701 | |
9563f88c YZ |
702 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
703 | struct ceph_cap *ocap, int mask); | |
de57606c SW |
704 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
705 | extern int __ceph_caps_used(struct ceph_inode_info *ci); | |
706 | ||
719a2514 YZ |
707 | static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci) |
708 | { | |
709 | return ci->i_nr_by_mode[0]; | |
710 | } | |
de57606c | 711 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); |
525d15e8 | 712 | extern int __ceph_caps_wanted(struct ceph_inode_info *ci); |
de57606c SW |
713 | |
714 | /* what the mds thinks we want */ | |
c1944fed | 715 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
de57606c | 716 | |
37151668 YS |
717 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
718 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); | |
fe33032d YZ |
719 | extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, |
720 | struct ceph_mount_options *fsopt); | |
e30ee581 | 721 | extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 722 | struct ceph_cap_reservation *ctx, int need); |
7bf8f736 | 723 | extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 724 | struct ceph_cap_reservation *ctx); |
3d14c5d2 | 725 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
de57606c | 726 | int *total, int *avail, int *used, |
85ccce43 | 727 | int *reserved, int *min); |
de57606c | 728 | |
de57606c | 729 | |
de57606c SW |
730 | |
731 | /* | |
732 | * we keep buffered readdir results attached to file->private_data | |
733 | */ | |
4918b6d1 | 734 | #define CEPH_F_SYNC 1 |
9cfa1098 | 735 | #define CEPH_F_ATEND 2 |
4918b6d1 | 736 | |
de57606c | 737 | struct ceph_file_info { |
252c6728 SW |
738 | short fmode; /* initialized on open */ |
739 | short flags; /* CEPH_F_* */ | |
de57606c | 740 | |
5d988308 YZ |
741 | spinlock_t rw_contexts_lock; |
742 | struct list_head rw_contexts; | |
f4b97866 YZ |
743 | |
744 | errseq_t meta_err; | |
81f148a9 | 745 | u32 filp_gen; |
ff5d913d | 746 | atomic_t num_locks; |
bb48bd4d CX |
747 | }; |
748 | ||
749 | struct ceph_dir_file_info { | |
750 | struct ceph_file_info file_info; | |
5d988308 | 751 | |
de57606c SW |
752 | /* readdir: position within the dir */ |
753 | u32 frag; | |
754 | struct ceph_mds_request *last_readdir; | |
de57606c SW |
755 | |
756 | /* readdir: position within a frag */ | |
f0494206 | 757 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
de57606c | 758 | char *last_name; /* last entry in previous chunk */ |
fdd4e158 YZ |
759 | long long dir_release_count; |
760 | long long dir_ordered_count; | |
761 | int readdir_cache_idx; | |
de57606c SW |
762 | |
763 | /* used for -o dirstat read() on directory thing */ | |
764 | char *dir_info; | |
765 | int dir_info_len; | |
766 | }; | |
767 | ||
5d988308 YZ |
768 | struct ceph_rw_context { |
769 | struct list_head list; | |
770 | struct task_struct *thread; | |
771 | int caps; | |
772 | }; | |
773 | ||
774 | #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ | |
775 | struct ceph_rw_context _name = { \ | |
776 | .thread = current, \ | |
777 | .caps = _caps, \ | |
778 | } | |
779 | ||
780 | static inline void ceph_add_rw_context(struct ceph_file_info *cf, | |
781 | struct ceph_rw_context *ctx) | |
782 | { | |
783 | spin_lock(&cf->rw_contexts_lock); | |
784 | list_add(&ctx->list, &cf->rw_contexts); | |
785 | spin_unlock(&cf->rw_contexts_lock); | |
786 | } | |
787 | ||
788 | static inline void ceph_del_rw_context(struct ceph_file_info *cf, | |
789 | struct ceph_rw_context *ctx) | |
790 | { | |
791 | spin_lock(&cf->rw_contexts_lock); | |
792 | list_del(&ctx->list); | |
793 | spin_unlock(&cf->rw_contexts_lock); | |
794 | } | |
795 | ||
796 | static inline struct ceph_rw_context* | |
797 | ceph_find_rw_context(struct ceph_file_info *cf) | |
798 | { | |
799 | struct ceph_rw_context *ctx, *found = NULL; | |
800 | spin_lock(&cf->rw_contexts_lock); | |
801 | list_for_each_entry(ctx, &cf->rw_contexts, list) { | |
802 | if (ctx->thread == current) { | |
803 | found = ctx; | |
804 | break; | |
805 | } | |
806 | } | |
807 | spin_unlock(&cf->rw_contexts_lock); | |
808 | return found; | |
809 | } | |
810 | ||
fdd4e158 YZ |
811 | struct ceph_readdir_cache_control { |
812 | struct page *page; | |
813 | struct dentry **dentries; | |
814 | int index; | |
815 | }; | |
de57606c | 816 | |
de57606c SW |
817 | /* |
818 | * A "snap realm" describes a subset of the file hierarchy sharing | |
819 | * the same set of snapshots that apply to it. The realms themselves | |
820 | * are organized into a hierarchy, such that children inherit (some of) | |
821 | * the snapshots of their parents. | |
822 | * | |
823 | * All inodes within the realm that have capabilities are linked into a | |
824 | * per-realm list. | |
825 | */ | |
826 | struct ceph_snap_realm { | |
827 | u64 ino; | |
e3161f17 | 828 | struct inode *inode; |
de57606c | 829 | atomic_t nref; |
a105f00c SW |
830 | struct rb_node node; |
831 | ||
de57606c SW |
832 | u64 created, seq; |
833 | u64 parent_ino; | |
834 | u64 parent_since; /* snapid when our current parent became so */ | |
835 | ||
836 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ | |
aa711ee3 | 837 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
de57606c | 838 | u64 *snaps; /* snaps specific to this realm */ |
aa711ee3 | 839 | u32 num_snaps; |
de57606c SW |
840 | |
841 | struct ceph_snap_realm *parent; | |
842 | struct list_head children; /* list of child realms */ | |
843 | struct list_head child_item; | |
844 | ||
845 | struct list_head empty_item; /* if i have ref==0 */ | |
846 | ||
ae00d4f3 SW |
847 | struct list_head dirty_item; /* if realm needs new context */ |
848 | ||
de57606c SW |
849 | /* the current set of snaps for this realm */ |
850 | struct ceph_snap_context *cached_context; | |
851 | ||
852 | struct list_head inodes_with_caps; | |
853 | spinlock_t inodes_with_caps_lock; | |
854 | }; | |
855 | ||
3d14c5d2 | 856 | static inline int default_congestion_kb(void) |
de57606c | 857 | { |
3d14c5d2 YS |
858 | int congestion_kb; |
859 | ||
860 | /* | |
861 | * Copied from NFS | |
862 | * | |
863 | * congestion size, scale with available memory. | |
864 | * | |
865 | * 64MB: 8192k | |
866 | * 128MB: 11585k | |
867 | * 256MB: 16384k | |
868 | * 512MB: 23170k | |
869 | * 1GB: 32768k | |
870 | * 2GB: 46340k | |
871 | * 4GB: 65536k | |
872 | * 8GB: 92681k | |
873 | * 16GB: 131072k | |
874 | * | |
875 | * This allows larger machines to have larger/more transfers. | |
876 | * Limit the default to 256M | |
877 | */ | |
ca79b0c2 | 878 | congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); |
3d14c5d2 YS |
879 | if (congestion_kb > 256*1024) |
880 | congestion_kb = 256*1024; | |
881 | ||
882 | return congestion_kb; | |
de57606c SW |
883 | } |
884 | ||
885 | ||
d468e729 YZ |
886 | /* super.c */ |
887 | extern int ceph_force_reconnect(struct super_block *sb); | |
de57606c SW |
888 | /* snap.c */ |
889 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, | |
890 | u64 ino); | |
891 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, | |
892 | struct ceph_snap_realm *realm); | |
893 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, | |
894 | struct ceph_snap_realm *realm); | |
895 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, | |
982d6011 YZ |
896 | void *p, void *e, bool deletion, |
897 | struct ceph_snap_realm **realm_ret); | |
de57606c | 898 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
2600d2dd | 899 | struct ceph_mds_session *session, |
de57606c | 900 | struct ceph_msg *msg); |
fc837c8f | 901 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
de57606c SW |
902 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
903 | struct ceph_cap_snap *capsnap); | |
904 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); | |
905 | ||
75c9627e YZ |
906 | extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, |
907 | u64 snap); | |
908 | extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, | |
909 | struct ceph_snapid_map *sm); | |
910 | extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); | |
911 | extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); | |
912 | ||
913 | ||
de57606c SW |
914 | /* |
915 | * a cap_snap is "pending" if it is still awaiting an in-progress | |
916 | * sync write (that may/may not still update size, mtime, etc.). | |
917 | */ | |
918 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) | |
919 | { | |
920 | return !list_empty(&ci->i_cap_snaps) && | |
86056090 YZ |
921 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
922 | ci_item)->writing; | |
de57606c SW |
923 | } |
924 | ||
de57606c | 925 | /* inode.c */ |
966c7160 JL |
926 | struct ceph_mds_reply_info_in; |
927 | struct ceph_mds_reply_dirfrag; | |
928 | ||
de57606c SW |
929 | extern const struct inode_operations ceph_file_iops; |
930 | ||
931 | extern struct inode *ceph_alloc_inode(struct super_block *sb); | |
87bc5b89 | 932 | extern void ceph_evict_inode(struct inode *inode); |
cfa6d412 | 933 | extern void ceph_free_inode(struct inode *inode); |
de57606c SW |
934 | |
935 | extern struct inode *ceph_get_inode(struct super_block *sb, | |
936 | struct ceph_vino vino); | |
937 | extern struct inode *ceph_get_snapdir(struct inode *parent); | |
938 | extern int ceph_fill_file_size(struct inode *inode, int issued, | |
939 | u32 truncate_seq, u64 truncate_size, u64 size); | |
940 | extern void ceph_fill_file_time(struct inode *inode, int issued, | |
9bbeab41 AB |
941 | u64 time_warp_seq, struct timespec64 *ctime, |
942 | struct timespec64 *mtime, | |
943 | struct timespec64 *atime); | |
966c7160 JL |
944 | extern int ceph_fill_inode(struct inode *inode, struct page *locked_page, |
945 | struct ceph_mds_reply_info_in *iinfo, | |
946 | struct ceph_mds_reply_dirfrag *dirinfo, | |
947 | struct ceph_mds_session *session, int cap_fmode, | |
948 | struct ceph_cap_reservation *caps_reservation); | |
de57606c | 949 | extern int ceph_fill_trace(struct super_block *sb, |
f5a03b08 | 950 | struct ceph_mds_request *req); |
de57606c SW |
951 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
952 | struct ceph_mds_session *session); | |
953 | ||
954 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); | |
955 | ||
efb0ca76 | 956 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
b415bf4f | 957 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
3c6f6b79 | 958 | extern void ceph_queue_vmtruncate(struct inode *inode); |
3c6f6b79 SW |
959 | extern void ceph_queue_invalidate(struct inode *inode); |
960 | extern void ceph_queue_writeback(struct inode *inode); | |
3e1d0452 | 961 | extern void ceph_async_iput(struct inode *inode); |
de57606c | 962 | |
01deead0 YZ |
963 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
964 | int mask, bool force); | |
965 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) | |
966 | { | |
967 | return __ceph_do_getattr(inode, NULL, mask, force); | |
968 | } | |
10556cb2 | 969 | extern int ceph_permission(struct inode *inode, int mask); |
a26fecca | 970 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
de57606c | 971 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
a528d35e DH |
972 | extern int ceph_getattr(const struct path *path, struct kstat *stat, |
973 | u32 request_mask, unsigned int flags); | |
de57606c SW |
974 | |
975 | /* xattr.c */ | |
a26fecca | 976 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
7221fe4c | 977 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
de57606c | 978 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
12fe3dda | 979 | extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
de57606c | 980 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); |
b1ee94aa | 981 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
de57606c | 982 | |
5c31e92d YZ |
983 | struct ceph_acl_sec_ctx { |
984 | #ifdef CONFIG_CEPH_FS_POSIX_ACL | |
985 | void *default_acl; | |
986 | void *acl; | |
ac6713cc YZ |
987 | #endif |
988 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL | |
989 | void *sec_ctx; | |
990 | u32 sec_ctxlen; | |
5c31e92d YZ |
991 | #endif |
992 | struct ceph_pagelist *pagelist; | |
993 | }; | |
994 | ||
315f2408 YZ |
995 | #ifdef CONFIG_SECURITY |
996 | extern bool ceph_security_xattr_deadlock(struct inode *in); | |
997 | extern bool ceph_security_xattr_wanted(struct inode *in); | |
998 | #else | |
999 | static inline bool ceph_security_xattr_deadlock(struct inode *in) | |
1000 | { | |
1001 | return false; | |
1002 | } | |
1003 | static inline bool ceph_security_xattr_wanted(struct inode *in) | |
1004 | { | |
1005 | return false; | |
1006 | } | |
1007 | #endif | |
1008 | ||
ac6713cc YZ |
1009 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL |
1010 | extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
1011 | struct ceph_acl_sec_ctx *ctx); | |
668959a5 JL |
1012 | static inline void ceph_security_invalidate_secctx(struct inode *inode) |
1013 | { | |
1014 | security_inode_invalidate_secctx(inode); | |
1015 | } | |
ac6713cc YZ |
1016 | #else |
1017 | static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
1018 | struct ceph_acl_sec_ctx *ctx) | |
1019 | { | |
1020 | return 0; | |
1021 | } | |
1022 | static inline void ceph_security_invalidate_secctx(struct inode *inode) | |
1023 | { | |
1024 | } | |
1025 | #endif | |
1026 | ||
5c31e92d | 1027 | void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); |
7221fe4c | 1028 | |
5c31e92d | 1029 | /* acl.c */ |
7221fe4c GZ |
1030 | #ifdef CONFIG_CEPH_FS_POSIX_ACL |
1031 | ||
1032 | struct posix_acl *ceph_get_acl(struct inode *, int); | |
72466d0b | 1033 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
b1ee94aa | 1034 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d YZ |
1035 | struct ceph_acl_sec_ctx *as_ctx); |
1036 | void ceph_init_inode_acls(struct inode *inode, | |
1037 | struct ceph_acl_sec_ctx *as_ctx); | |
c969d9bf GZ |
1038 | |
1039 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
1040 | { | |
1041 | forget_all_cached_acls(inode); | |
1042 | } | |
7221fe4c GZ |
1043 | |
1044 | #else | |
1045 | ||
1046 | #define ceph_get_acl NULL | |
72466d0b | 1047 | #define ceph_set_acl NULL |
7221fe4c | 1048 | |
b1ee94aa | 1049 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d | 1050 | struct ceph_acl_sec_ctx *as_ctx) |
7221fe4c GZ |
1051 | { |
1052 | return 0; | |
1053 | } | |
b1ee94aa | 1054 | static inline void ceph_init_inode_acls(struct inode *inode, |
5c31e92d | 1055 | struct ceph_acl_sec_ctx *as_ctx) |
b1ee94aa YZ |
1056 | { |
1057 | } | |
7221fe4c GZ |
1058 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
1059 | { | |
1060 | return 0; | |
1061 | } | |
1062 | ||
1063 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
1064 | { | |
1065 | } | |
1066 | ||
1067 | #endif | |
1068 | ||
de57606c SW |
1069 | /* caps.c */ |
1070 | extern const char *ceph_cap_string(int c); | |
1071 | extern void ceph_handle_caps(struct ceph_mds_session *session, | |
1072 | struct ceph_msg *msg); | |
d9df2783 YZ |
1073 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
1074 | struct ceph_cap_reservation *ctx); | |
1075 | extern void ceph_add_cap(struct inode *inode, | |
1076 | struct ceph_mds_session *session, u64 cap_id, | |
135e671e | 1077 | unsigned issued, unsigned wanted, |
d9df2783 YZ |
1078 | unsigned cap, unsigned seq, u64 realmino, int flags, |
1079 | struct ceph_cap **new_cap); | |
a096b09a | 1080 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
d6e47819 | 1081 | extern void __ceph_remove_caps(struct ceph_inode_info *ci); |
37151668 YS |
1082 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
1083 | struct ceph_cap *cap); | |
9215aeea | 1084 | extern int ceph_is_any_caps(struct inode *inode); |
de57606c | 1085 | |
f1a3d572 | 1086 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
02c24a82 JB |
1087 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
1088 | int datasync); | |
e548e9b9 YZ |
1089 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1090 | struct ceph_mds_session *session); | |
de57606c SW |
1091 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1092 | struct ceph_mds_session *session); | |
e8a4d267 JL |
1093 | void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, |
1094 | struct ceph_inode_info *ci); | |
2bc50259 GF |
1095 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
1096 | int mds); | |
40dcf75e JL |
1097 | extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, |
1098 | bool snap_rwsem_locked); | |
de57606c SW |
1099 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); |
1100 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); | |
e64f44a8 XL |
1101 | extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, |
1102 | int had); | |
de57606c SW |
1103 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, |
1104 | struct ceph_snap_context *snapc); | |
ed9b430c YZ |
1105 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
1106 | struct ceph_mds_session **psession); | |
efb0ca76 | 1107 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
de57606c SW |
1108 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
1109 | struct ceph_mds_session *session); | |
afcdaea3 SW |
1110 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
1111 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); | |
6ef0bc6d | 1112 | extern int ceph_drop_caps_for_unlink(struct inode *inode); |
de57606c SW |
1113 | extern int ceph_encode_inode_release(void **p, struct inode *inode, |
1114 | int mds, int drop, int unless, int force); | |
1115 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, | |
ca6c8ae0 | 1116 | struct inode *dir, |
de57606c SW |
1117 | int mds, int drop, int unless); |
1118 | ||
5e3ded1b | 1119 | extern int ceph_get_caps(struct file *filp, int need, int want, |
31c542a1 | 1120 | loff_t endoff, int *got, struct page **pinned_page); |
5e3ded1b | 1121 | extern int ceph_try_get_caps(struct inode *inode, |
2ee9dd95 | 1122 | int need, int want, bool nonblock, int *got); |
de57606c SW |
1123 | |
1124 | /* for counting open files by mode */ | |
719a2514 YZ |
1125 | extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count); |
1126 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count); | |
1127 | extern void __ceph_touch_fmode(struct ceph_inode_info *ci, | |
1128 | struct ceph_mds_client *mdsc, int fmode); | |
de57606c SW |
1129 | |
1130 | /* addr.c */ | |
1131 | extern const struct address_space_operations ceph_aops; | |
1132 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); | |
10183a69 | 1133 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
5e3ded1b | 1134 | extern int ceph_pool_perm_check(struct inode *inode, int need); |
10183a69 | 1135 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); |
de57606c SW |
1136 | |
1137 | /* file.c */ | |
1138 | extern const struct file_operations ceph_file_fops; | |
9e0eb85d | 1139 | |
719a2514 | 1140 | extern int ceph_renew_caps(struct inode *inode, int fmode); |
de57606c | 1141 | extern int ceph_open(struct inode *inode, struct file *file); |
5ef50c3b | 1142 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
44907d79 | 1143 | struct file *file, unsigned flags, umode_t mode); |
de57606c | 1144 | extern int ceph_release(struct inode *inode, struct file *filp); |
31c542a1 YZ |
1145 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
1146 | char *data, size_t len); | |
55f2a045 | 1147 | |
de57606c SW |
1148 | /* dir.c */ |
1149 | extern const struct file_operations ceph_dir_fops; | |
38c48b5f | 1150 | extern const struct file_operations ceph_snapdir_fops; |
de57606c | 1151 | extern const struct inode_operations ceph_dir_iops; |
38c48b5f | 1152 | extern const struct inode_operations ceph_snapdir_iops; |
18fc8abd | 1153 | extern const struct dentry_operations ceph_dentry_ops; |
de57606c | 1154 | |
f3c4ebe6 | 1155 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
de57606c | 1156 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
468640e3 SW |
1157 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
1158 | struct dentry *dentry, int err); | |
de57606c SW |
1159 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
1160 | struct dentry *dentry, int err); | |
1161 | ||
37c4efc1 YZ |
1162 | extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); |
1163 | extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); | |
81a6cf2d | 1164 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
37c4efc1 | 1165 | extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); |
e5f86dc3 | 1166 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
fdd4e158 | 1167 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
de57606c | 1168 | |
de57606c SW |
1169 | /* ioctl.c */ |
1170 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); | |
1171 | ||
1172 | /* export.c */ | |
1173 | extern const struct export_operations ceph_export_ops; | |
3886274a | 1174 | struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); |
de57606c | 1175 | |
40819f6f | 1176 | /* locks.c */ |
eb13e832 | 1177 | extern __init void ceph_flock_init(void); |
40819f6f GF |
1178 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
1179 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); | |
1180 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); | |
39be95e9 JS |
1181 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
1182 | struct ceph_filelock *flocks, | |
1183 | int num_fcntl_locks, | |
1184 | int num_flock_locks); | |
1185 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, | |
1186 | struct ceph_pagelist *pagelist, | |
1187 | int num_fcntl_locks, int num_flock_locks); | |
40819f6f | 1188 | |
3d14c5d2 | 1189 | /* debugfs.c */ |
1a829ff2 | 1190 | extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); |
3d14c5d2 YS |
1191 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); |
1192 | ||
fb18a575 | 1193 | /* quota.c */ |
d557c48d LH |
1194 | static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) |
1195 | { | |
1196 | return ci->i_max_files || ci->i_max_bytes; | |
1197 | } | |
1198 | ||
1199 | extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); | |
1200 | ||
1201 | static inline void __ceph_update_quota(struct ceph_inode_info *ci, | |
1202 | u64 max_bytes, u64 max_files) | |
1203 | { | |
1204 | bool had_quota, has_quota; | |
1205 | had_quota = __ceph_has_any_quota(ci); | |
1206 | ci->i_max_bytes = max_bytes; | |
1207 | ci->i_max_files = max_files; | |
1208 | has_quota = __ceph_has_any_quota(ci); | |
1209 | ||
1210 | if (had_quota != has_quota) | |
1211 | ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); | |
1212 | } | |
1213 | ||
fb18a575 LH |
1214 | extern void ceph_handle_quota(struct ceph_mds_client *mdsc, |
1215 | struct ceph_mds_session *session, | |
1216 | struct ceph_msg *msg); | |
b7a29217 | 1217 | extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); |
2b83845f LH |
1218 | extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, |
1219 | loff_t newlen); | |
1ab302a0 LH |
1220 | extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, |
1221 | loff_t newlen); | |
9122eed5 LH |
1222 | extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, |
1223 | struct kstatfs *buf); | |
dffdcd71 LH |
1224 | extern int ceph_quota_check_rename(struct ceph_mds_client *mdsc, |
1225 | struct inode *old, struct inode *new); | |
0c44a8e0 | 1226 | extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); |
fb18a575 | 1227 | |
de57606c | 1228 | #endif /* _FS_CEPH_SUPER_H */ |