]>
Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
de57606c SW |
2 | #ifndef _FS_CEPH_SUPER_H |
3 | #define _FS_CEPH_SUPER_H | |
4 | ||
3d14c5d2 | 5 | #include <linux/ceph/ceph_debug.h> |
de57606c SW |
6 | |
7 | #include <asm/unaligned.h> | |
8 | #include <linux/backing-dev.h> | |
9 | #include <linux/completion.h> | |
10 | #include <linux/exportfs.h> | |
11 | #include <linux/fs.h> | |
12 | #include <linux/mempool.h> | |
13 | #include <linux/pagemap.h> | |
14 | #include <linux/wait.h> | |
f1a3d572 | 15 | #include <linux/writeback.h> |
5a0e3ad6 | 16 | #include <linux/slab.h> |
c969d9bf | 17 | #include <linux/posix_acl.h> |
805692d0 | 18 | #include <linux/refcount.h> |
de57606c | 19 | |
3d14c5d2 | 20 | #include <linux/ceph/libceph.h> |
de57606c | 21 | |
99ccbd22 MT |
22 | #ifdef CONFIG_CEPH_FSCACHE |
23 | #include <linux/fscache.h> | |
24 | #endif | |
25 | ||
de57606c SW |
26 | /* f_type in struct statfs */ |
27 | #define CEPH_SUPER_MAGIC 0x00c36400 | |
28 | ||
29 | /* large granularity for statfs utilization stats to facilitate | |
30 | * large volume sizes on 32-bit machines. */ | |
92a49fb0 | 31 | #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ |
de57606c SW |
32 | #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) |
33 | ||
3d14c5d2 YS |
34 | #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ |
35 | #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ | |
36 | #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ | |
ad1fee96 | 37 | #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ |
a40dc6cc | 38 | #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ |
99ccbd22 | 39 | #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ |
10183a69 | 40 | #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ |
e9e427f0 | 41 | #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ |
9122eed5 | 42 | #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ |
ea4cdc54 | 43 | #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ |
6a259382 | 44 | |
6f9718fe LH |
45 | #define CEPH_MOUNT_OPT_DEFAULT \ |
46 | (CEPH_MOUNT_OPT_DCACHE | \ | |
47 | CEPH_MOUNT_OPT_NOCOPYFROM) | |
de57606c | 48 | |
3d14c5d2 YS |
49 | #define ceph_set_mount_opt(fsc, opt) \ |
50 | (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; | |
51 | #define ceph_test_mount_opt(fsc, opt) \ | |
52 | (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) | |
de57606c | 53 | |
aa187926 YZ |
54 | /* max size of osd read request, limited by libceph */ |
55 | #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN | |
95cca2b4 YZ |
56 | /* osd has a configurable limitaion of max write size. |
57 | * CEPH_MSG_MAX_DATA_LEN should be small enough. */ | |
58 | #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN | |
7c94ba27 | 59 | #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ |
3d14c5d2 YS |
60 | #define CEPH_MAX_READDIR_DEFAULT 1024 |
61 | #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) | |
62 | #define CEPH_SNAPDIRNAME_DEFAULT ".snap" | |
de57606c | 63 | |
4214fb15 YZ |
64 | /* |
65 | * Delay telling the MDS we no longer want caps, in case we reopen | |
66 | * the file. Delay a minimum amount of time, even if we send a cap | |
67 | * message for some other reason. Otherwise, take the oppotunity to | |
68 | * update the mds to avoid sending another message later. | |
69 | */ | |
70 | #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ | |
71 | #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ | |
72 | ||
3d14c5d2 | 73 | struct ceph_mount_options { |
6e19a16e | 74 | int flags; |
3d14c5d2 YS |
75 | int sb_flags; |
76 | ||
83817e35 SW |
77 | int wsize; /* max write size */ |
78 | int rsize; /* max read size */ | |
79 | int rasize; /* max readahead */ | |
6e19a16e SW |
80 | int congestion_kb; /* max writeback in flight */ |
81 | int caps_wanted_delay_min, caps_wanted_delay_max; | |
23804d91 SW |
82 | int max_readdir; /* max readdir result (entires) */ |
83 | int max_readdir_bytes; /* max readdir result (bytes) */ | |
de57606c | 84 | |
3d14c5d2 YS |
85 | /* |
86 | * everything above this point can be memcmp'd; everything below | |
87 | * is handled in compare_mount_options() | |
88 | */ | |
de57606c | 89 | |
3d14c5d2 | 90 | char *snapdir_name; /* default ".snap" */ |
430afbad | 91 | char *mds_namespace; /* default NULL */ |
3f384954 | 92 | char *server_path; /* default "/" */ |
1d8f8360 | 93 | char *fscache_uniq; /* default NULL */ |
3d14c5d2 | 94 | }; |
de57606c | 95 | |
3d14c5d2 | 96 | struct ceph_fs_client { |
de57606c SW |
97 | struct super_block *sb; |
98 | ||
3d14c5d2 YS |
99 | struct ceph_mount_options *mount_options; |
100 | struct ceph_client *client; | |
de57606c | 101 | |
3d14c5d2 | 102 | unsigned long mount_state; |
85ccce43 | 103 | int min_caps; /* min caps i added */ |
719784ba | 104 | loff_t max_file_size; |
85ccce43 | 105 | |
3d14c5d2 | 106 | struct ceph_mds_client *mdsc; |
de57606c SW |
107 | |
108 | /* writeback */ | |
109 | mempool_t *wb_pagevec_pool; | |
e3ec8d68 YZ |
110 | atomic_long_t writeback_count; |
111 | ||
de57606c SW |
112 | struct workqueue_struct *wb_wq; |
113 | struct workqueue_struct *pg_inv_wq; | |
114 | struct workqueue_struct *trunc_wq; | |
e3ec8d68 | 115 | struct workqueue_struct *cap_wq; |
de57606c | 116 | |
0743304d | 117 | #ifdef CONFIG_DEBUG_FS |
3d14c5d2 | 118 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
2baba250 | 119 | struct dentry *debugfs_congestion_kb; |
06edf046 | 120 | struct dentry *debugfs_bdi; |
3d14c5d2 | 121 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
14ed9703 | 122 | struct dentry *debugfs_mds_sessions; |
0743304d | 123 | #endif |
99ccbd22 MT |
124 | |
125 | #ifdef CONFIG_CEPH_FSCACHE | |
126 | struct fscache_cookie *fscache; | |
99ccbd22 | 127 | #endif |
de57606c SW |
128 | }; |
129 | ||
3d14c5d2 | 130 | |
de57606c SW |
131 | /* |
132 | * File i/o capability. This tracks shared state with the metadata | |
133 | * server that allows us to cache or writeback attributes or to read | |
134 | * and write data. For any given inode, we should have one or more | |
135 | * capabilities, one issued by each metadata server, and our | |
136 | * cumulative access is the OR of all issued capabilities. | |
137 | * | |
138 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds | |
139 | * session capability lists. | |
140 | */ | |
141 | struct ceph_cap { | |
142 | struct ceph_inode_info *ci; | |
143 | struct rb_node ci_node; /* per-ci cap tree */ | |
144 | struct ceph_mds_session *session; | |
145 | struct list_head session_caps; /* per-session caplist */ | |
de57606c | 146 | u64 cap_id; /* unique cap id (mds provided) */ |
745a8e3b YZ |
147 | union { |
148 | /* in-use caps */ | |
149 | struct { | |
150 | int issued; /* latest, from the mds */ | |
151 | int implemented; /* implemented superset of | |
152 | issued (for revocation) */ | |
153 | int mds, mds_wanted; | |
154 | }; | |
155 | /* caps to release */ | |
156 | struct { | |
157 | u64 cap_ino; | |
158 | int queue_release; | |
159 | }; | |
160 | }; | |
685f9a5d SW |
161 | u32 seq, issue_seq, mseq; |
162 | u32 cap_gen; /* active/stale cycle */ | |
de57606c SW |
163 | unsigned long last_used; |
164 | struct list_head caps_item; | |
165 | }; | |
166 | ||
167 | #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ | |
168 | #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ | |
169 | #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ | |
170 | ||
0e294387 YZ |
171 | struct ceph_cap_flush { |
172 | u64 tid; | |
173 | int caps; /* 0 means capsnap */ | |
c8799fc4 | 174 | bool wake; /* wake up flush waiters when finish ? */ |
0e294387 YZ |
175 | struct list_head g_list; // global |
176 | struct list_head i_list; // per inode | |
177 | }; | |
178 | ||
de57606c SW |
179 | /* |
180 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, | |
181 | * we first complete any in-process sync writes and writeback any dirty | |
182 | * data before flushing the snapped state (tracked here) back to the MDS. | |
183 | */ | |
184 | struct ceph_cap_snap { | |
805692d0 | 185 | refcount_t nref; |
0e294387 | 186 | struct list_head ci_item; |
de57606c | 187 | |
0e294387 YZ |
188 | struct ceph_cap_flush cap_flush; |
189 | ||
190 | u64 follows; | |
de57606c SW |
191 | int issued, dirty; |
192 | struct ceph_snap_context *context; | |
193 | ||
5706b27d | 194 | umode_t mode; |
05cb11c1 EB |
195 | kuid_t uid; |
196 | kgid_t gid; | |
de57606c | 197 | |
4a625be4 | 198 | struct ceph_buffer *xattr_blob; |
de57606c SW |
199 | u64 xattr_version; |
200 | ||
201 | u64 size; | |
9bbeab41 | 202 | struct timespec64 mtime, atime, ctime; |
de57606c | 203 | u64 time_warp_seq; |
5f743e45 YZ |
204 | u64 truncate_size; |
205 | u32 truncate_seq; | |
de57606c SW |
206 | int writing; /* a sync write is still in progress */ |
207 | int dirty_pages; /* dirty pages awaiting writeback */ | |
e20d258d | 208 | bool inline_data; |
86056090 | 209 | bool need_flush; |
de57606c SW |
210 | }; |
211 | ||
212 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) | |
213 | { | |
805692d0 | 214 | if (refcount_dec_and_test(&capsnap->nref)) { |
4a625be4 SW |
215 | if (capsnap->xattr_blob) |
216 | ceph_buffer_put(capsnap->xattr_blob); | |
de57606c | 217 | kfree(capsnap); |
4a625be4 | 218 | } |
de57606c SW |
219 | } |
220 | ||
221 | /* | |
222 | * The frag tree describes how a directory is fragmented, potentially across | |
223 | * multiple metadata servers. It is also used to indicate points where | |
224 | * metadata authority is delegated, and whether/where metadata is replicated. | |
225 | * | |
226 | * A _leaf_ frag will be present in the i_fragtree IFF there is | |
227 | * delegation info. That is, if mds >= 0 || ndist > 0. | |
228 | */ | |
229 | #define CEPH_MAX_DIRFRAG_REP 4 | |
230 | ||
231 | struct ceph_inode_frag { | |
232 | struct rb_node node; | |
233 | ||
234 | /* fragtree state */ | |
235 | u32 frag; | |
236 | int split_by; /* i.e. 2^(split_by) children */ | |
237 | ||
238 | /* delegation and replication info */ | |
239 | int mds; /* -1 if same authority as parent */ | |
240 | int ndist; /* >0 if replicated */ | |
241 | int dist[CEPH_MAX_DIRFRAG_REP]; | |
242 | }; | |
243 | ||
244 | /* | |
245 | * We cache inode xattrs as an encoded blob until they are first used, | |
246 | * at which point we parse them into an rbtree. | |
247 | */ | |
248 | struct ceph_inode_xattr { | |
249 | struct rb_node node; | |
250 | ||
251 | const char *name; | |
252 | int name_len; | |
253 | const char *val; | |
254 | int val_len; | |
255 | int dirty; | |
256 | ||
257 | int should_free_name; | |
258 | int should_free_val; | |
259 | }; | |
260 | ||
3d14c5d2 YS |
261 | /* |
262 | * Ceph dentry state | |
263 | */ | |
264 | struct ceph_dentry_info { | |
37c4efc1 | 265 | struct dentry *dentry; |
3d14c5d2 | 266 | struct ceph_mds_session *lease_session; |
37c4efc1 YZ |
267 | struct list_head lease_list; |
268 | unsigned flags; | |
97aeb6bf YZ |
269 | int lease_shared_gen; |
270 | u32 lease_gen; | |
3d14c5d2 YS |
271 | u32 lease_seq; |
272 | unsigned long lease_renew_after, lease_renew_from; | |
9b16f03c | 273 | unsigned long time; |
3d14c5d2 YS |
274 | u64 offset; |
275 | }; | |
276 | ||
37c4efc1 YZ |
277 | #define CEPH_DENTRY_REFERENCED 1 |
278 | #define CEPH_DENTRY_LEASE_LIST 2 | |
279 | #define CEPH_DENTRY_SHRINK_LIST 4 | |
280 | ||
de57606c SW |
281 | struct ceph_inode_xattrs_info { |
282 | /* | |
283 | * (still encoded) xattr blob. we avoid the overhead of parsing | |
284 | * this until someone actually calls getxattr, etc. | |
285 | * | |
286 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == | |
287 | * NULL means we don't know. | |
288 | */ | |
289 | struct ceph_buffer *blob, *prealloc_blob; | |
290 | ||
291 | struct rb_root index; | |
292 | bool dirty; | |
293 | int count; | |
294 | int names_size; | |
295 | int vals_size; | |
296 | u64 version, index_version; | |
297 | }; | |
298 | ||
299 | /* | |
300 | * Ceph inode. | |
301 | */ | |
de57606c SW |
302 | struct ceph_inode_info { |
303 | struct ceph_vino i_vino; /* ceph ino + snap */ | |
304 | ||
be655596 SW |
305 | spinlock_t i_ceph_lock; |
306 | ||
de57606c | 307 | u64 i_version; |
31c542a1 | 308 | u64 i_inline_version; |
de57606c SW |
309 | u32 i_time_warp_seq; |
310 | ||
311 | unsigned i_ceph_flags; | |
fdd4e158 YZ |
312 | atomic64_t i_release_count; |
313 | atomic64_t i_ordered_count; | |
314 | atomic64_t i_complete_seq[2]; | |
de57606c | 315 | |
6c0f3af7 | 316 | struct ceph_dir_layout i_dir_layout; |
de57606c SW |
317 | struct ceph_file_layout i_layout; |
318 | char *i_symlink; | |
319 | ||
320 | /* for dirs */ | |
9bbeab41 | 321 | struct timespec64 i_rctime; |
de57606c SW |
322 | u64 i_rbytes, i_rfiles, i_rsubdirs; |
323 | u64 i_files, i_subdirs; | |
de57606c | 324 | |
fb18a575 LH |
325 | /* quotas */ |
326 | u64 i_max_bytes, i_max_files; | |
327 | ||
08796873 YZ |
328 | s32 i_dir_pin; |
329 | ||
de57606c | 330 | struct rb_root i_fragtree; |
1b1bc16d | 331 | int i_fragtree_nsplits; |
de57606c SW |
332 | struct mutex i_fragtree_mutex; |
333 | ||
334 | struct ceph_inode_xattrs_info i_xattrs; | |
335 | ||
be655596 | 336 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
de57606c SW |
337 | * s_mutex. */ |
338 | struct rb_root i_caps; /* cap list */ | |
339 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ | |
340 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ | |
341 | struct list_head i_dirty_item, i_flushing_item; | |
de57606c SW |
342 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
343 | * overlapping, pipelined cap flushes to the mds. we can probably | |
344 | * reduce the tid to 8 bits if we're concerned about inode size. */ | |
f66fd9f0 | 345 | struct ceph_cap_flush *i_prealloc_cap_flush; |
e4500b5e | 346 | struct list_head i_cap_flush_list; |
de57606c SW |
347 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
348 | unsigned long i_hold_caps_min; /* jiffies */ | |
349 | unsigned long i_hold_caps_max; /* jiffies */ | |
350 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ | |
de57606c SW |
351 | struct ceph_cap_reservation i_cap_migration_resv; |
352 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ | |
7d8cb26d SW |
353 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
354 | dirty|flushing caps */ | |
de57606c SW |
355 | unsigned i_snap_caps; /* cap bits for snapped files */ |
356 | ||
774a6a11 | 357 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
de57606c | 358 | |
b0d7c223 | 359 | struct mutex i_truncate_mutex; |
de57606c SW |
360 | u32 i_truncate_seq; /* last truncate to smaller size */ |
361 | u64 i_truncate_size; /* and the size we last truncated down to */ | |
362 | int i_truncate_pending; /* still need to call vmtruncate */ | |
363 | ||
364 | u64 i_max_size; /* max file size authorized by mds */ | |
365 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ | |
366 | u64 i_wanted_max_size; /* offset we'd like to write too */ | |
367 | u64 i_requested_max_size; /* max_size we've requested */ | |
368 | ||
369 | /* held references to caps */ | |
370 | int i_pin_ref; | |
d3d0720d | 371 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; |
de57606c | 372 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
89aa5930 | 373 | atomic_t i_filelock_ref; |
97aeb6bf | 374 | atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ |
cd045cb4 | 375 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
de57606c SW |
376 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
377 | ||
de57606c | 378 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
68cd5b4b | 379 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
de57606c SW |
380 | spinlock_t i_unsafe_lock; |
381 | ||
75c9627e YZ |
382 | union { |
383 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ | |
384 | struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ | |
385 | }; | |
de57606c SW |
386 | int i_snap_realm_counter; /* snap realm (if caps) */ |
387 | struct list_head i_snap_realm_item; | |
388 | struct list_head i_snap_flush_item; | |
389 | ||
390 | struct work_struct i_wb_work; /* writeback work */ | |
391 | struct work_struct i_pg_inv_work; /* page invalidation work */ | |
392 | ||
393 | struct work_struct i_vmtruncate_work; | |
394 | ||
99ccbd22 MT |
395 | #ifdef CONFIG_CEPH_FSCACHE |
396 | struct fscache_cookie *fscache; | |
f7f7e7a0 | 397 | u32 i_fscache_gen; |
99ccbd22 | 398 | #endif |
de57606c SW |
399 | struct inode vfs_inode; /* at end */ |
400 | }; | |
401 | ||
402 | static inline struct ceph_inode_info *ceph_inode(struct inode *inode) | |
403 | { | |
fbbccec9 | 404 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
de57606c SW |
405 | } |
406 | ||
ad1fee96 YS |
407 | static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) |
408 | { | |
409 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; | |
410 | } | |
411 | ||
412 | static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) | |
413 | { | |
414 | return (struct ceph_fs_client *)sb->s_fs_info; | |
415 | } | |
416 | ||
3d14c5d2 YS |
417 | static inline struct ceph_vino ceph_vino(struct inode *inode) |
418 | { | |
419 | return ceph_inode(inode)->i_vino; | |
420 | } | |
421 | ||
422 | /* | |
423 | * ino_t is <64 bits on many architectures, blech. | |
424 | * | |
ad1fee96 YS |
425 | * i_ino (kernel inode) st_ino (userspace) |
426 | * i386 32 32 | |
427 | * x86_64+ino32 64 32 | |
428 | * x86_64 64 64 | |
429 | */ | |
3310f754 | 430 | static inline u32 ceph_ino_to_ino32(__u64 vino) |
ad1fee96 | 431 | { |
3310f754 AO |
432 | u32 ino = vino & 0xffffffff; |
433 | ino ^= vino >> 32; | |
ad1fee96 | 434 | if (!ino) |
a661fc56 | 435 | ino = 2; |
ad1fee96 YS |
436 | return ino; |
437 | } | |
438 | ||
439 | /* | |
440 | * kernel i_ino value | |
3d14c5d2 YS |
441 | */ |
442 | static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) | |
443 | { | |
3d14c5d2 | 444 | #if BITS_PER_LONG == 32 |
3310f754 AO |
445 | return ceph_ino_to_ino32(vino.ino); |
446 | #else | |
447 | return (ino_t)vino.ino; | |
3d14c5d2 | 448 | #endif |
3d14c5d2 YS |
449 | } |
450 | ||
ad1fee96 YS |
451 | /* |
452 | * user-visible ino (stat, filldir) | |
453 | */ | |
454 | #if BITS_PER_LONG == 32 | |
455 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
456 | { | |
457 | return ino; | |
458 | } | |
459 | #else | |
460 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
461 | { | |
462 | if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) | |
463 | ino = ceph_ino_to_ino32(ino); | |
464 | return ino; | |
465 | } | |
466 | #endif | |
467 | ||
468 | ||
3d14c5d2 YS |
469 | /* for printf-style formatting */ |
470 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap | |
471 | ||
472 | static inline u64 ceph_ino(struct inode *inode) | |
473 | { | |
474 | return ceph_inode(inode)->i_vino.ino; | |
475 | } | |
476 | static inline u64 ceph_snap(struct inode *inode) | |
477 | { | |
478 | return ceph_inode(inode)->i_vino.snap; | |
479 | } | |
480 | ||
481 | static inline int ceph_ino_compare(struct inode *inode, void *data) | |
482 | { | |
483 | struct ceph_vino *pvino = (struct ceph_vino *)data; | |
484 | struct ceph_inode_info *ci = ceph_inode(inode); | |
485 | return ci->i_vino.ino == pvino->ino && | |
486 | ci->i_vino.snap == pvino->snap; | |
487 | } | |
488 | ||
489 | static inline struct inode *ceph_find_inode(struct super_block *sb, | |
490 | struct ceph_vino vino) | |
491 | { | |
492 | ino_t t = ceph_vino_to_ino(vino); | |
493 | return ilookup5(sb, t, ceph_ino_compare, &vino); | |
494 | } | |
495 | ||
496 | ||
497 | /* | |
498 | * Ceph inode. | |
499 | */ | |
10183a69 YZ |
500 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
501 | #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ | |
502 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ | |
503 | #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ | |
504 | #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ | |
505 | #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ | |
506 | #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ | |
315f2408 | 507 | #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ |
77310320 | 508 | #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ |
13c2b57d | 509 | #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ |
70220ac8 | 510 | #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ |
26544c62 | 511 | #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ |
b3f8d68f YZ |
512 | #define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */ |
513 | ||
26544c62 JL |
514 | |
515 | /* | |
516 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode | |
517 | * and then clear it when they start succeeding. Note that we do a lockless | |
518 | * check first, and only take the lock if it looks like it needs to be changed. | |
519 | * The write submission code just takes this as a hint, so we're not too | |
520 | * worried if a few slip through in either direction. | |
521 | */ | |
522 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) | |
523 | { | |
524 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { | |
525 | spin_lock(&ci->i_ceph_lock); | |
526 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; | |
527 | spin_unlock(&ci->i_ceph_lock); | |
528 | } | |
529 | } | |
530 | ||
531 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) | |
532 | { | |
533 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { | |
534 | spin_lock(&ci->i_ceph_lock); | |
535 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; | |
536 | spin_unlock(&ci->i_ceph_lock); | |
537 | } | |
538 | } | |
3d14c5d2 | 539 | |
2f276c51 | 540 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
fdd4e158 YZ |
541 | long long release_count, |
542 | long long ordered_count) | |
de57606c | 543 | { |
fdd4e158 YZ |
544 | smp_mb__before_atomic(); |
545 | atomic64_set(&ci->i_complete_seq[0], release_count); | |
546 | atomic64_set(&ci->i_complete_seq[1], ordered_count); | |
de57606c SW |
547 | } |
548 | ||
2f276c51 | 549 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
de57606c | 550 | { |
fdd4e158 YZ |
551 | atomic64_inc(&ci->i_release_count); |
552 | } | |
553 | ||
554 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) | |
555 | { | |
556 | atomic64_inc(&ci->i_ordered_count); | |
2f276c51 | 557 | } |
de57606c | 558 | |
2f276c51 YZ |
559 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
560 | { | |
fdd4e158 YZ |
561 | return atomic64_read(&ci->i_complete_seq[0]) == |
562 | atomic64_read(&ci->i_release_count); | |
de57606c SW |
563 | } |
564 | ||
70db4f36 YZ |
565 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
566 | { | |
fdd4e158 YZ |
567 | return atomic64_read(&ci->i_complete_seq[0]) == |
568 | atomic64_read(&ci->i_release_count) && | |
569 | atomic64_read(&ci->i_complete_seq[1]) == | |
570 | atomic64_read(&ci->i_ordered_count); | |
70db4f36 YZ |
571 | } |
572 | ||
2f276c51 | 573 | static inline void ceph_dir_clear_complete(struct inode *inode) |
de57606c | 574 | { |
2f276c51 YZ |
575 | __ceph_dir_clear_complete(ceph_inode(inode)); |
576 | } | |
de57606c | 577 | |
70db4f36 | 578 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
2f276c51 | 579 | { |
fdd4e158 | 580 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
de57606c SW |
581 | } |
582 | ||
70db4f36 YZ |
583 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
584 | { | |
fdd4e158 YZ |
585 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
586 | smp_rmb(); | |
70db4f36 YZ |
587 | return ret; |
588 | } | |
de57606c SW |
589 | |
590 | /* find a specific frag @f */ | |
591 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, | |
592 | u32 f); | |
593 | ||
594 | /* | |
595 | * choose fragment for value @v. copy frag content to pfrag, if leaf | |
596 | * exists | |
597 | */ | |
598 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, | |
599 | struct ceph_inode_frag *pfrag, | |
600 | int *found); | |
601 | ||
1e9c2eb6 | 602 | static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) |
de57606c SW |
603 | { |
604 | return (struct ceph_dentry_info *)dentry->d_fsdata; | |
605 | } | |
606 | ||
de57606c SW |
607 | /* |
608 | * caps helpers | |
609 | */ | |
610 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) | |
611 | { | |
612 | return !RB_EMPTY_ROOT(&ci->i_caps); | |
613 | } | |
614 | ||
615 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); | |
616 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); | |
617 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, | |
618 | struct ceph_cap *cap); | |
619 | ||
620 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) | |
621 | { | |
622 | int issued; | |
be655596 | 623 | spin_lock(&ci->i_ceph_lock); |
de57606c | 624 | issued = __ceph_caps_issued(ci, NULL); |
be655596 | 625 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
626 | return issued; |
627 | } | |
628 | ||
629 | static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, | |
630 | int touch) | |
631 | { | |
632 | int r; | |
be655596 | 633 | spin_lock(&ci->i_ceph_lock); |
de57606c | 634 | r = __ceph_caps_issued_mask(ci, mask, touch); |
be655596 | 635 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
636 | return r; |
637 | } | |
638 | ||
639 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) | |
640 | { | |
641 | return ci->i_dirty_caps | ci->i_flushing_caps; | |
642 | } | |
f66fd9f0 YZ |
643 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
644 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); | |
645 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, | |
646 | struct ceph_cap_flush **pcf); | |
de57606c | 647 | |
9563f88c YZ |
648 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
649 | struct ceph_cap *ocap, int mask); | |
de57606c SW |
650 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
651 | extern int __ceph_caps_used(struct ceph_inode_info *ci); | |
652 | ||
653 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); | |
654 | ||
655 | /* | |
656 | * wanted, by virtue of open file modes AND cap refs (buffered/cached data) | |
657 | */ | |
658 | static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) | |
659 | { | |
660 | int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); | |
661 | if (w & CEPH_CAP_FILE_BUFFER) | |
662 | w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ | |
663 | return w; | |
664 | } | |
665 | ||
666 | /* what the mds thinks we want */ | |
c1944fed | 667 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
de57606c | 668 | |
37151668 YS |
669 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
670 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); | |
671 | extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); | |
e30ee581 | 672 | extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 673 | struct ceph_cap_reservation *ctx, int need); |
7bf8f736 | 674 | extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 675 | struct ceph_cap_reservation *ctx); |
3d14c5d2 | 676 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
de57606c | 677 | int *total, int *avail, int *used, |
85ccce43 | 678 | int *reserved, int *min); |
de57606c | 679 | |
de57606c | 680 | |
de57606c SW |
681 | |
682 | /* | |
683 | * we keep buffered readdir results attached to file->private_data | |
684 | */ | |
4918b6d1 | 685 | #define CEPH_F_SYNC 1 |
9cfa1098 | 686 | #define CEPH_F_ATEND 2 |
4918b6d1 | 687 | |
de57606c | 688 | struct ceph_file_info { |
252c6728 SW |
689 | short fmode; /* initialized on open */ |
690 | short flags; /* CEPH_F_* */ | |
de57606c | 691 | |
5d988308 YZ |
692 | spinlock_t rw_contexts_lock; |
693 | struct list_head rw_contexts; | |
bb48bd4d CX |
694 | }; |
695 | ||
696 | struct ceph_dir_file_info { | |
697 | struct ceph_file_info file_info; | |
5d988308 | 698 | |
de57606c SW |
699 | /* readdir: position within the dir */ |
700 | u32 frag; | |
701 | struct ceph_mds_request *last_readdir; | |
de57606c SW |
702 | |
703 | /* readdir: position within a frag */ | |
f0494206 | 704 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
de57606c | 705 | char *last_name; /* last entry in previous chunk */ |
fdd4e158 YZ |
706 | long long dir_release_count; |
707 | long long dir_ordered_count; | |
708 | int readdir_cache_idx; | |
de57606c SW |
709 | |
710 | /* used for -o dirstat read() on directory thing */ | |
711 | char *dir_info; | |
712 | int dir_info_len; | |
713 | }; | |
714 | ||
5d988308 YZ |
715 | struct ceph_rw_context { |
716 | struct list_head list; | |
717 | struct task_struct *thread; | |
718 | int caps; | |
719 | }; | |
720 | ||
721 | #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ | |
722 | struct ceph_rw_context _name = { \ | |
723 | .thread = current, \ | |
724 | .caps = _caps, \ | |
725 | } | |
726 | ||
727 | static inline void ceph_add_rw_context(struct ceph_file_info *cf, | |
728 | struct ceph_rw_context *ctx) | |
729 | { | |
730 | spin_lock(&cf->rw_contexts_lock); | |
731 | list_add(&ctx->list, &cf->rw_contexts); | |
732 | spin_unlock(&cf->rw_contexts_lock); | |
733 | } | |
734 | ||
735 | static inline void ceph_del_rw_context(struct ceph_file_info *cf, | |
736 | struct ceph_rw_context *ctx) | |
737 | { | |
738 | spin_lock(&cf->rw_contexts_lock); | |
739 | list_del(&ctx->list); | |
740 | spin_unlock(&cf->rw_contexts_lock); | |
741 | } | |
742 | ||
743 | static inline struct ceph_rw_context* | |
744 | ceph_find_rw_context(struct ceph_file_info *cf) | |
745 | { | |
746 | struct ceph_rw_context *ctx, *found = NULL; | |
747 | spin_lock(&cf->rw_contexts_lock); | |
748 | list_for_each_entry(ctx, &cf->rw_contexts, list) { | |
749 | if (ctx->thread == current) { | |
750 | found = ctx; | |
751 | break; | |
752 | } | |
753 | } | |
754 | spin_unlock(&cf->rw_contexts_lock); | |
755 | return found; | |
756 | } | |
757 | ||
fdd4e158 YZ |
758 | struct ceph_readdir_cache_control { |
759 | struct page *page; | |
760 | struct dentry **dentries; | |
761 | int index; | |
762 | }; | |
de57606c | 763 | |
de57606c SW |
764 | /* |
765 | * A "snap realm" describes a subset of the file hierarchy sharing | |
766 | * the same set of snapshots that apply to it. The realms themselves | |
767 | * are organized into a hierarchy, such that children inherit (some of) | |
768 | * the snapshots of their parents. | |
769 | * | |
770 | * All inodes within the realm that have capabilities are linked into a | |
771 | * per-realm list. | |
772 | */ | |
773 | struct ceph_snap_realm { | |
774 | u64 ino; | |
e3161f17 | 775 | struct inode *inode; |
de57606c | 776 | atomic_t nref; |
a105f00c SW |
777 | struct rb_node node; |
778 | ||
de57606c SW |
779 | u64 created, seq; |
780 | u64 parent_ino; | |
781 | u64 parent_since; /* snapid when our current parent became so */ | |
782 | ||
783 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ | |
aa711ee3 | 784 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
de57606c | 785 | u64 *snaps; /* snaps specific to this realm */ |
aa711ee3 | 786 | u32 num_snaps; |
de57606c SW |
787 | |
788 | struct ceph_snap_realm *parent; | |
789 | struct list_head children; /* list of child realms */ | |
790 | struct list_head child_item; | |
791 | ||
792 | struct list_head empty_item; /* if i have ref==0 */ | |
793 | ||
ae00d4f3 SW |
794 | struct list_head dirty_item; /* if realm needs new context */ |
795 | ||
de57606c SW |
796 | /* the current set of snaps for this realm */ |
797 | struct ceph_snap_context *cached_context; | |
798 | ||
799 | struct list_head inodes_with_caps; | |
800 | spinlock_t inodes_with_caps_lock; | |
801 | }; | |
802 | ||
3d14c5d2 | 803 | static inline int default_congestion_kb(void) |
de57606c | 804 | { |
3d14c5d2 YS |
805 | int congestion_kb; |
806 | ||
807 | /* | |
808 | * Copied from NFS | |
809 | * | |
810 | * congestion size, scale with available memory. | |
811 | * | |
812 | * 64MB: 8192k | |
813 | * 128MB: 11585k | |
814 | * 256MB: 16384k | |
815 | * 512MB: 23170k | |
816 | * 1GB: 32768k | |
817 | * 2GB: 46340k | |
818 | * 4GB: 65536k | |
819 | * 8GB: 92681k | |
820 | * 16GB: 131072k | |
821 | * | |
822 | * This allows larger machines to have larger/more transfers. | |
823 | * Limit the default to 256M | |
824 | */ | |
ca79b0c2 | 825 | congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); |
3d14c5d2 YS |
826 | if (congestion_kb > 256*1024) |
827 | congestion_kb = 256*1024; | |
828 | ||
829 | return congestion_kb; | |
de57606c SW |
830 | } |
831 | ||
832 | ||
833 | ||
834 | /* snap.c */ | |
835 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, | |
836 | u64 ino); | |
837 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, | |
838 | struct ceph_snap_realm *realm); | |
839 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, | |
840 | struct ceph_snap_realm *realm); | |
841 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, | |
982d6011 YZ |
842 | void *p, void *e, bool deletion, |
843 | struct ceph_snap_realm **realm_ret); | |
de57606c | 844 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
2600d2dd | 845 | struct ceph_mds_session *session, |
de57606c | 846 | struct ceph_msg *msg); |
fc837c8f | 847 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
de57606c SW |
848 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
849 | struct ceph_cap_snap *capsnap); | |
850 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); | |
851 | ||
75c9627e YZ |
852 | extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, |
853 | u64 snap); | |
854 | extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, | |
855 | struct ceph_snapid_map *sm); | |
856 | extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); | |
857 | extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); | |
858 | ||
859 | ||
de57606c SW |
860 | /* |
861 | * a cap_snap is "pending" if it is still awaiting an in-progress | |
862 | * sync write (that may/may not still update size, mtime, etc.). | |
863 | */ | |
864 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) | |
865 | { | |
866 | return !list_empty(&ci->i_cap_snaps) && | |
86056090 YZ |
867 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
868 | ci_item)->writing; | |
de57606c SW |
869 | } |
870 | ||
de57606c SW |
871 | /* inode.c */ |
872 | extern const struct inode_operations ceph_file_iops; | |
873 | ||
874 | extern struct inode *ceph_alloc_inode(struct super_block *sb); | |
875 | extern void ceph_destroy_inode(struct inode *inode); | |
9f12bd11 | 876 | extern int ceph_drop_inode(struct inode *inode); |
de57606c SW |
877 | |
878 | extern struct inode *ceph_get_inode(struct super_block *sb, | |
879 | struct ceph_vino vino); | |
880 | extern struct inode *ceph_get_snapdir(struct inode *parent); | |
881 | extern int ceph_fill_file_size(struct inode *inode, int issued, | |
882 | u32 truncate_seq, u64 truncate_size, u64 size); | |
883 | extern void ceph_fill_file_time(struct inode *inode, int issued, | |
9bbeab41 AB |
884 | u64 time_warp_seq, struct timespec64 *ctime, |
885 | struct timespec64 *mtime, | |
886 | struct timespec64 *atime); | |
de57606c | 887 | extern int ceph_fill_trace(struct super_block *sb, |
f5a03b08 | 888 | struct ceph_mds_request *req); |
de57606c SW |
889 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
890 | struct ceph_mds_session *session); | |
891 | ||
892 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); | |
893 | ||
efb0ca76 | 894 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
b415bf4f | 895 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
3c6f6b79 SW |
896 | extern void ceph_queue_vmtruncate(struct inode *inode); |
897 | ||
898 | extern void ceph_queue_invalidate(struct inode *inode); | |
899 | extern void ceph_queue_writeback(struct inode *inode); | |
de57606c | 900 | |
01deead0 YZ |
901 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
902 | int mask, bool force); | |
903 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) | |
904 | { | |
905 | return __ceph_do_getattr(inode, NULL, mask, force); | |
906 | } | |
10556cb2 | 907 | extern int ceph_permission(struct inode *inode, int mask); |
a26fecca | 908 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
de57606c | 909 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
a528d35e DH |
910 | extern int ceph_getattr(const struct path *path, struct kstat *stat, |
911 | u32 request_mask, unsigned int flags); | |
de57606c SW |
912 | |
913 | /* xattr.c */ | |
a26fecca | 914 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
7221fe4c | 915 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
de57606c | 916 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
de57606c SW |
917 | extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
918 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); | |
3ce6cd12 AE |
919 | extern void __init ceph_xattr_init(void); |
920 | extern void ceph_xattr_exit(void); | |
b1ee94aa | 921 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
de57606c | 922 | |
315f2408 YZ |
923 | #ifdef CONFIG_SECURITY |
924 | extern bool ceph_security_xattr_deadlock(struct inode *in); | |
925 | extern bool ceph_security_xattr_wanted(struct inode *in); | |
926 | #else | |
927 | static inline bool ceph_security_xattr_deadlock(struct inode *in) | |
928 | { | |
929 | return false; | |
930 | } | |
931 | static inline bool ceph_security_xattr_wanted(struct inode *in) | |
932 | { | |
933 | return false; | |
934 | } | |
935 | #endif | |
936 | ||
7221fe4c | 937 | /* acl.c */ |
b1ee94aa YZ |
938 | struct ceph_acls_info { |
939 | void *default_acl; | |
940 | void *acl; | |
941 | struct ceph_pagelist *pagelist; | |
942 | }; | |
7221fe4c GZ |
943 | |
944 | #ifdef CONFIG_CEPH_FS_POSIX_ACL | |
945 | ||
946 | struct posix_acl *ceph_get_acl(struct inode *, int); | |
72466d0b | 947 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
b1ee94aa YZ |
948 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
949 | struct ceph_acls_info *info); | |
950 | void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); | |
951 | void ceph_release_acls_info(struct ceph_acls_info *info); | |
c969d9bf GZ |
952 | |
953 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
954 | { | |
955 | forget_all_cached_acls(inode); | |
956 | } | |
7221fe4c GZ |
957 | |
958 | #else | |
959 | ||
960 | #define ceph_get_acl NULL | |
72466d0b | 961 | #define ceph_set_acl NULL |
7221fe4c | 962 | |
b1ee94aa YZ |
963 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
964 | struct ceph_acls_info *info) | |
7221fe4c GZ |
965 | { |
966 | return 0; | |
967 | } | |
b1ee94aa YZ |
968 | static inline void ceph_init_inode_acls(struct inode *inode, |
969 | struct ceph_acls_info *info) | |
970 | { | |
971 | } | |
972 | static inline void ceph_release_acls_info(struct ceph_acls_info *info) | |
973 | { | |
974 | } | |
7221fe4c GZ |
975 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
976 | { | |
977 | return 0; | |
978 | } | |
979 | ||
980 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
981 | { | |
982 | } | |
983 | ||
984 | #endif | |
985 | ||
de57606c SW |
986 | /* caps.c */ |
987 | extern const char *ceph_cap_string(int c); | |
988 | extern void ceph_handle_caps(struct ceph_mds_session *session, | |
989 | struct ceph_msg *msg); | |
d9df2783 YZ |
990 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
991 | struct ceph_cap_reservation *ctx); | |
992 | extern void ceph_add_cap(struct inode *inode, | |
993 | struct ceph_mds_session *session, u64 cap_id, | |
994 | int fmode, unsigned issued, unsigned wanted, | |
995 | unsigned cap, unsigned seq, u64 realmino, int flags, | |
996 | struct ceph_cap **new_cap); | |
a096b09a | 997 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
e3ec8d68 | 998 | extern void __ceph_remove_caps(struct inode* inode); |
37151668 YS |
999 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
1000 | struct ceph_cap *cap); | |
9215aeea | 1001 | extern int ceph_is_any_caps(struct inode *inode); |
de57606c | 1002 | |
f1a3d572 | 1003 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
02c24a82 JB |
1004 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
1005 | int datasync); | |
e548e9b9 YZ |
1006 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1007 | struct ceph_mds_session *session); | |
de57606c SW |
1008 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1009 | struct ceph_mds_session *session); | |
2bc50259 GF |
1010 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
1011 | int mds); | |
de57606c SW |
1012 | extern int ceph_get_cap_mds(struct inode *inode); |
1013 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); | |
1014 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); | |
1015 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, | |
1016 | struct ceph_snap_context *snapc); | |
ed9b430c YZ |
1017 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
1018 | struct ceph_mds_session **psession); | |
efb0ca76 | 1019 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
de57606c SW |
1020 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
1021 | struct ceph_mds_session *session); | |
afcdaea3 SW |
1022 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
1023 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); | |
6ef0bc6d | 1024 | extern int ceph_drop_caps_for_unlink(struct inode *inode); |
de57606c SW |
1025 | extern int ceph_encode_inode_release(void **p, struct inode *inode, |
1026 | int mds, int drop, int unless, int force); | |
1027 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, | |
ca6c8ae0 | 1028 | struct inode *dir, |
de57606c SW |
1029 | int mds, int drop, int unless); |
1030 | ||
1031 | extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, | |
31c542a1 | 1032 | loff_t endoff, int *got, struct page **pinned_page); |
2b1ac852 | 1033 | extern int ceph_try_get_caps(struct ceph_inode_info *ci, |
2ee9dd95 | 1034 | int need, int want, bool nonblock, int *got); |
de57606c SW |
1035 | |
1036 | /* for counting open files by mode */ | |
774a6a11 | 1037 | extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); |
de57606c SW |
1038 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); |
1039 | ||
1040 | /* addr.c */ | |
1041 | extern const struct address_space_operations ceph_aops; | |
1042 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); | |
10183a69 YZ |
1043 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
1044 | extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); | |
1045 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); | |
de57606c SW |
1046 | |
1047 | /* file.c */ | |
1048 | extern const struct file_operations ceph_file_fops; | |
9e0eb85d | 1049 | |
77310320 | 1050 | extern int ceph_renew_caps(struct inode *inode); |
de57606c | 1051 | extern int ceph_open(struct inode *inode, struct file *file); |
5ef50c3b | 1052 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
44907d79 | 1053 | struct file *file, unsigned flags, umode_t mode); |
de57606c | 1054 | extern int ceph_release(struct inode *inode, struct file *filp); |
31c542a1 YZ |
1055 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
1056 | char *data, size_t len); | |
55f2a045 | 1057 | |
de57606c SW |
1058 | /* dir.c */ |
1059 | extern const struct file_operations ceph_dir_fops; | |
38c48b5f | 1060 | extern const struct file_operations ceph_snapdir_fops; |
de57606c | 1061 | extern const struct inode_operations ceph_dir_iops; |
38c48b5f | 1062 | extern const struct inode_operations ceph_snapdir_iops; |
18fc8abd | 1063 | extern const struct dentry_operations ceph_dentry_ops; |
de57606c | 1064 | |
f3c4ebe6 | 1065 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
de57606c | 1066 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
468640e3 SW |
1067 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
1068 | struct dentry *dentry, int err); | |
de57606c SW |
1069 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
1070 | struct dentry *dentry, int err); | |
1071 | ||
37c4efc1 YZ |
1072 | extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); |
1073 | extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); | |
81a6cf2d | 1074 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
37c4efc1 | 1075 | extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); |
e5f86dc3 | 1076 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
fdd4e158 | 1077 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
de57606c | 1078 | |
de57606c SW |
1079 | /* ioctl.c */ |
1080 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); | |
1081 | ||
1082 | /* export.c */ | |
1083 | extern const struct export_operations ceph_export_ops; | |
1084 | ||
40819f6f | 1085 | /* locks.c */ |
eb13e832 | 1086 | extern __init void ceph_flock_init(void); |
40819f6f GF |
1087 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
1088 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); | |
1089 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); | |
39be95e9 JS |
1090 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
1091 | struct ceph_filelock *flocks, | |
1092 | int num_fcntl_locks, | |
1093 | int num_flock_locks); | |
1094 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, | |
1095 | struct ceph_pagelist *pagelist, | |
1096 | int num_fcntl_locks, int num_flock_locks); | |
40819f6f | 1097 | |
3d14c5d2 YS |
1098 | /* debugfs.c */ |
1099 | extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); | |
1100 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); | |
1101 | ||
fb18a575 | 1102 | /* quota.c */ |
d557c48d LH |
1103 | static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) |
1104 | { | |
1105 | return ci->i_max_files || ci->i_max_bytes; | |
1106 | } | |
1107 | ||
1108 | extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); | |
1109 | ||
1110 | static inline void __ceph_update_quota(struct ceph_inode_info *ci, | |
1111 | u64 max_bytes, u64 max_files) | |
1112 | { | |
1113 | bool had_quota, has_quota; | |
1114 | had_quota = __ceph_has_any_quota(ci); | |
1115 | ci->i_max_bytes = max_bytes; | |
1116 | ci->i_max_files = max_files; | |
1117 | has_quota = __ceph_has_any_quota(ci); | |
1118 | ||
1119 | if (had_quota != has_quota) | |
1120 | ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); | |
1121 | } | |
1122 | ||
fb18a575 LH |
1123 | extern void ceph_handle_quota(struct ceph_mds_client *mdsc, |
1124 | struct ceph_mds_session *session, | |
1125 | struct ceph_msg *msg); | |
b7a29217 | 1126 | extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); |
cafe21a4 | 1127 | extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); |
2b83845f LH |
1128 | extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, |
1129 | loff_t newlen); | |
1ab302a0 LH |
1130 | extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, |
1131 | loff_t newlen); | |
9122eed5 LH |
1132 | extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, |
1133 | struct kstatfs *buf); | |
fb18a575 | 1134 | |
de57606c | 1135 | #endif /* _FS_CEPH_SUPER_H */ |