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; | |
fe33032d | 82 | int caps_max; |
23804d91 SW |
83 | int max_readdir; /* max readdir result (entires) */ |
84 | int max_readdir_bytes; /* max readdir result (bytes) */ | |
de57606c | 85 | |
3d14c5d2 YS |
86 | /* |
87 | * everything above this point can be memcmp'd; everything below | |
88 | * is handled in compare_mount_options() | |
89 | */ | |
de57606c | 90 | |
3d14c5d2 | 91 | char *snapdir_name; /* default ".snap" */ |
430afbad | 92 | char *mds_namespace; /* default NULL */ |
3f384954 | 93 | char *server_path; /* default "/" */ |
1d8f8360 | 94 | char *fscache_uniq; /* default NULL */ |
3d14c5d2 | 95 | }; |
de57606c | 96 | |
3d14c5d2 | 97 | struct ceph_fs_client { |
de57606c SW |
98 | struct super_block *sb; |
99 | ||
3d14c5d2 YS |
100 | struct ceph_mount_options *mount_options; |
101 | struct ceph_client *client; | |
de57606c | 102 | |
3d14c5d2 | 103 | unsigned long mount_state; |
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 | ||
1cf89a8d | 112 | struct workqueue_struct *inode_wq; |
e3ec8d68 | 113 | struct workqueue_struct *cap_wq; |
de57606c | 114 | |
0743304d | 115 | #ifdef CONFIG_DEBUG_FS |
3d14c5d2 | 116 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
2baba250 | 117 | struct dentry *debugfs_congestion_kb; |
06edf046 | 118 | struct dentry *debugfs_bdi; |
3d14c5d2 | 119 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
14ed9703 | 120 | struct dentry *debugfs_mds_sessions; |
0743304d | 121 | #endif |
99ccbd22 MT |
122 | |
123 | #ifdef CONFIG_CEPH_FSCACHE | |
124 | struct fscache_cookie *fscache; | |
99ccbd22 | 125 | #endif |
de57606c SW |
126 | }; |
127 | ||
3d14c5d2 | 128 | |
de57606c SW |
129 | /* |
130 | * File i/o capability. This tracks shared state with the metadata | |
131 | * server that allows us to cache or writeback attributes or to read | |
132 | * and write data. For any given inode, we should have one or more | |
133 | * capabilities, one issued by each metadata server, and our | |
134 | * cumulative access is the OR of all issued capabilities. | |
135 | * | |
136 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds | |
137 | * session capability lists. | |
138 | */ | |
139 | struct ceph_cap { | |
140 | struct ceph_inode_info *ci; | |
141 | struct rb_node ci_node; /* per-ci cap tree */ | |
142 | struct ceph_mds_session *session; | |
143 | struct list_head session_caps; /* per-session caplist */ | |
de57606c | 144 | u64 cap_id; /* unique cap id (mds provided) */ |
745a8e3b YZ |
145 | union { |
146 | /* in-use caps */ | |
147 | struct { | |
148 | int issued; /* latest, from the mds */ | |
149 | int implemented; /* implemented superset of | |
150 | issued (for revocation) */ | |
151 | int mds, mds_wanted; | |
152 | }; | |
153 | /* caps to release */ | |
154 | struct { | |
155 | u64 cap_ino; | |
156 | int queue_release; | |
157 | }; | |
158 | }; | |
685f9a5d SW |
159 | u32 seq, issue_seq, mseq; |
160 | u32 cap_gen; /* active/stale cycle */ | |
de57606c SW |
161 | unsigned long last_used; |
162 | struct list_head caps_item; | |
163 | }; | |
164 | ||
165 | #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ | |
166 | #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ | |
167 | #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ | |
168 | ||
0e294387 YZ |
169 | struct ceph_cap_flush { |
170 | u64 tid; | |
171 | int caps; /* 0 means capsnap */ | |
c8799fc4 | 172 | bool wake; /* wake up flush waiters when finish ? */ |
0e294387 YZ |
173 | struct list_head g_list; // global |
174 | struct list_head i_list; // per inode | |
175 | }; | |
176 | ||
de57606c SW |
177 | /* |
178 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, | |
179 | * we first complete any in-process sync writes and writeback any dirty | |
180 | * data before flushing the snapped state (tracked here) back to the MDS. | |
181 | */ | |
182 | struct ceph_cap_snap { | |
805692d0 | 183 | refcount_t nref; |
0e294387 | 184 | struct list_head ci_item; |
de57606c | 185 | |
0e294387 YZ |
186 | struct ceph_cap_flush cap_flush; |
187 | ||
188 | u64 follows; | |
de57606c SW |
189 | int issued, dirty; |
190 | struct ceph_snap_context *context; | |
191 | ||
5706b27d | 192 | umode_t mode; |
05cb11c1 EB |
193 | kuid_t uid; |
194 | kgid_t gid; | |
de57606c | 195 | |
4a625be4 | 196 | struct ceph_buffer *xattr_blob; |
de57606c SW |
197 | u64 xattr_version; |
198 | ||
199 | u64 size; | |
9bbeab41 | 200 | struct timespec64 mtime, atime, ctime; |
de57606c | 201 | u64 time_warp_seq; |
5f743e45 YZ |
202 | u64 truncate_size; |
203 | u32 truncate_seq; | |
de57606c SW |
204 | int writing; /* a sync write is still in progress */ |
205 | int dirty_pages; /* dirty pages awaiting writeback */ | |
e20d258d | 206 | bool inline_data; |
86056090 | 207 | bool need_flush; |
de57606c SW |
208 | }; |
209 | ||
210 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) | |
211 | { | |
805692d0 | 212 | if (refcount_dec_and_test(&capsnap->nref)) { |
4a625be4 SW |
213 | if (capsnap->xattr_blob) |
214 | ceph_buffer_put(capsnap->xattr_blob); | |
de57606c | 215 | kfree(capsnap); |
4a625be4 | 216 | } |
de57606c SW |
217 | } |
218 | ||
219 | /* | |
220 | * The frag tree describes how a directory is fragmented, potentially across | |
221 | * multiple metadata servers. It is also used to indicate points where | |
222 | * metadata authority is delegated, and whether/where metadata is replicated. | |
223 | * | |
224 | * A _leaf_ frag will be present in the i_fragtree IFF there is | |
225 | * delegation info. That is, if mds >= 0 || ndist > 0. | |
226 | */ | |
227 | #define CEPH_MAX_DIRFRAG_REP 4 | |
228 | ||
229 | struct ceph_inode_frag { | |
230 | struct rb_node node; | |
231 | ||
232 | /* fragtree state */ | |
233 | u32 frag; | |
234 | int split_by; /* i.e. 2^(split_by) children */ | |
235 | ||
236 | /* delegation and replication info */ | |
237 | int mds; /* -1 if same authority as parent */ | |
238 | int ndist; /* >0 if replicated */ | |
239 | int dist[CEPH_MAX_DIRFRAG_REP]; | |
240 | }; | |
241 | ||
242 | /* | |
243 | * We cache inode xattrs as an encoded blob until they are first used, | |
244 | * at which point we parse them into an rbtree. | |
245 | */ | |
246 | struct ceph_inode_xattr { | |
247 | struct rb_node node; | |
248 | ||
249 | const char *name; | |
250 | int name_len; | |
251 | const char *val; | |
252 | int val_len; | |
253 | int dirty; | |
254 | ||
255 | int should_free_name; | |
256 | int should_free_val; | |
257 | }; | |
258 | ||
3d14c5d2 YS |
259 | /* |
260 | * Ceph dentry state | |
261 | */ | |
262 | struct ceph_dentry_info { | |
37c4efc1 | 263 | struct dentry *dentry; |
3d14c5d2 | 264 | struct ceph_mds_session *lease_session; |
37c4efc1 YZ |
265 | struct list_head lease_list; |
266 | unsigned flags; | |
97aeb6bf YZ |
267 | int lease_shared_gen; |
268 | u32 lease_gen; | |
3d14c5d2 YS |
269 | u32 lease_seq; |
270 | unsigned long lease_renew_after, lease_renew_from; | |
9b16f03c | 271 | unsigned long time; |
3d14c5d2 YS |
272 | u64 offset; |
273 | }; | |
274 | ||
37c4efc1 YZ |
275 | #define CEPH_DENTRY_REFERENCED 1 |
276 | #define CEPH_DENTRY_LEASE_LIST 2 | |
277 | #define CEPH_DENTRY_SHRINK_LIST 4 | |
278 | ||
de57606c SW |
279 | struct ceph_inode_xattrs_info { |
280 | /* | |
281 | * (still encoded) xattr blob. we avoid the overhead of parsing | |
282 | * this until someone actually calls getxattr, etc. | |
283 | * | |
284 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == | |
285 | * NULL means we don't know. | |
286 | */ | |
287 | struct ceph_buffer *blob, *prealloc_blob; | |
288 | ||
289 | struct rb_root index; | |
290 | bool dirty; | |
291 | int count; | |
292 | int names_size; | |
293 | int vals_size; | |
294 | u64 version, index_version; | |
295 | }; | |
296 | ||
297 | /* | |
298 | * Ceph inode. | |
299 | */ | |
de57606c SW |
300 | struct ceph_inode_info { |
301 | struct ceph_vino i_vino; /* ceph ino + snap */ | |
302 | ||
be655596 SW |
303 | spinlock_t i_ceph_lock; |
304 | ||
de57606c | 305 | u64 i_version; |
31c542a1 | 306 | u64 i_inline_version; |
de57606c SW |
307 | u32 i_time_warp_seq; |
308 | ||
309 | unsigned i_ceph_flags; | |
fdd4e158 YZ |
310 | atomic64_t i_release_count; |
311 | atomic64_t i_ordered_count; | |
312 | atomic64_t i_complete_seq[2]; | |
de57606c | 313 | |
6c0f3af7 | 314 | struct ceph_dir_layout i_dir_layout; |
de57606c SW |
315 | struct ceph_file_layout i_layout; |
316 | char *i_symlink; | |
317 | ||
318 | /* for dirs */ | |
9bbeab41 | 319 | struct timespec64 i_rctime; |
de57606c SW |
320 | u64 i_rbytes, i_rfiles, i_rsubdirs; |
321 | u64 i_files, i_subdirs; | |
de57606c | 322 | |
fb18a575 LH |
323 | /* quotas */ |
324 | u64 i_max_bytes, i_max_files; | |
325 | ||
08796873 YZ |
326 | s32 i_dir_pin; |
327 | ||
de57606c | 328 | struct rb_root i_fragtree; |
1b1bc16d | 329 | int i_fragtree_nsplits; |
de57606c SW |
330 | struct mutex i_fragtree_mutex; |
331 | ||
332 | struct ceph_inode_xattrs_info i_xattrs; | |
333 | ||
be655596 | 334 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
de57606c SW |
335 | * s_mutex. */ |
336 | struct rb_root i_caps; /* cap list */ | |
337 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ | |
338 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ | |
339 | struct list_head i_dirty_item, i_flushing_item; | |
de57606c SW |
340 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
341 | * overlapping, pipelined cap flushes to the mds. we can probably | |
342 | * reduce the tid to 8 bits if we're concerned about inode size. */ | |
f66fd9f0 | 343 | struct ceph_cap_flush *i_prealloc_cap_flush; |
e4500b5e | 344 | struct list_head i_cap_flush_list; |
de57606c SW |
345 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
346 | unsigned long i_hold_caps_min; /* jiffies */ | |
347 | unsigned long i_hold_caps_max; /* jiffies */ | |
348 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ | |
de57606c SW |
349 | struct ceph_cap_reservation i_cap_migration_resv; |
350 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ | |
7d8cb26d SW |
351 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
352 | dirty|flushing caps */ | |
de57606c SW |
353 | unsigned i_snap_caps; /* cap bits for snapped files */ |
354 | ||
774a6a11 | 355 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
de57606c | 356 | |
b0d7c223 | 357 | struct mutex i_truncate_mutex; |
de57606c SW |
358 | u32 i_truncate_seq; /* last truncate to smaller size */ |
359 | u64 i_truncate_size; /* and the size we last truncated down to */ | |
360 | int i_truncate_pending; /* still need to call vmtruncate */ | |
361 | ||
362 | u64 i_max_size; /* max file size authorized by mds */ | |
363 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ | |
364 | u64 i_wanted_max_size; /* offset we'd like to write too */ | |
365 | u64 i_requested_max_size; /* max_size we've requested */ | |
366 | ||
367 | /* held references to caps */ | |
368 | int i_pin_ref; | |
d3d0720d | 369 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; |
de57606c | 370 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
89aa5930 | 371 | atomic_t i_filelock_ref; |
97aeb6bf | 372 | atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ |
cd045cb4 | 373 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
de57606c SW |
374 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
375 | ||
de57606c | 376 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
68cd5b4b | 377 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
de57606c SW |
378 | spinlock_t i_unsafe_lock; |
379 | ||
75c9627e YZ |
380 | union { |
381 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ | |
382 | struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ | |
383 | }; | |
de57606c SW |
384 | int i_snap_realm_counter; /* snap realm (if caps) */ |
385 | struct list_head i_snap_realm_item; | |
386 | struct list_head i_snap_flush_item; | |
245ce991 | 387 | struct timespec64 i_btime; |
193e7b37 | 388 | struct timespec64 i_snap_btime; |
de57606c | 389 | |
1cf89a8d YZ |
390 | struct work_struct i_work; |
391 | unsigned long i_work_mask; | |
de57606c | 392 | |
99ccbd22 MT |
393 | #ifdef CONFIG_CEPH_FSCACHE |
394 | struct fscache_cookie *fscache; | |
f7f7e7a0 | 395 | u32 i_fscache_gen; |
99ccbd22 | 396 | #endif |
de57606c SW |
397 | struct inode vfs_inode; /* at end */ |
398 | }; | |
399 | ||
400 | static inline struct ceph_inode_info *ceph_inode(struct inode *inode) | |
401 | { | |
fbbccec9 | 402 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
de57606c SW |
403 | } |
404 | ||
ad1fee96 YS |
405 | static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) |
406 | { | |
407 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; | |
408 | } | |
409 | ||
410 | static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) | |
411 | { | |
412 | return (struct ceph_fs_client *)sb->s_fs_info; | |
413 | } | |
414 | ||
3d14c5d2 YS |
415 | static inline struct ceph_vino ceph_vino(struct inode *inode) |
416 | { | |
417 | return ceph_inode(inode)->i_vino; | |
418 | } | |
419 | ||
420 | /* | |
421 | * ino_t is <64 bits on many architectures, blech. | |
422 | * | |
ad1fee96 YS |
423 | * i_ino (kernel inode) st_ino (userspace) |
424 | * i386 32 32 | |
425 | * x86_64+ino32 64 32 | |
426 | * x86_64 64 64 | |
427 | */ | |
3310f754 | 428 | static inline u32 ceph_ino_to_ino32(__u64 vino) |
ad1fee96 | 429 | { |
3310f754 AO |
430 | u32 ino = vino & 0xffffffff; |
431 | ino ^= vino >> 32; | |
ad1fee96 | 432 | if (!ino) |
a661fc56 | 433 | ino = 2; |
ad1fee96 YS |
434 | return ino; |
435 | } | |
436 | ||
437 | /* | |
438 | * kernel i_ino value | |
3d14c5d2 YS |
439 | */ |
440 | static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) | |
441 | { | |
3d14c5d2 | 442 | #if BITS_PER_LONG == 32 |
3310f754 AO |
443 | return ceph_ino_to_ino32(vino.ino); |
444 | #else | |
445 | return (ino_t)vino.ino; | |
3d14c5d2 | 446 | #endif |
3d14c5d2 YS |
447 | } |
448 | ||
ad1fee96 YS |
449 | /* |
450 | * user-visible ino (stat, filldir) | |
451 | */ | |
452 | #if BITS_PER_LONG == 32 | |
453 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
454 | { | |
455 | return ino; | |
456 | } | |
457 | #else | |
458 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
459 | { | |
460 | if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) | |
461 | ino = ceph_ino_to_ino32(ino); | |
462 | return ino; | |
463 | } | |
464 | #endif | |
465 | ||
466 | ||
3d14c5d2 YS |
467 | /* for printf-style formatting */ |
468 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap | |
469 | ||
470 | static inline u64 ceph_ino(struct inode *inode) | |
471 | { | |
472 | return ceph_inode(inode)->i_vino.ino; | |
473 | } | |
474 | static inline u64 ceph_snap(struct inode *inode) | |
475 | { | |
476 | return ceph_inode(inode)->i_vino.snap; | |
477 | } | |
478 | ||
479 | static inline int ceph_ino_compare(struct inode *inode, void *data) | |
480 | { | |
481 | struct ceph_vino *pvino = (struct ceph_vino *)data; | |
482 | struct ceph_inode_info *ci = ceph_inode(inode); | |
483 | return ci->i_vino.ino == pvino->ino && | |
484 | ci->i_vino.snap == pvino->snap; | |
485 | } | |
486 | ||
487 | static inline struct inode *ceph_find_inode(struct super_block *sb, | |
488 | struct ceph_vino vino) | |
489 | { | |
490 | ino_t t = ceph_vino_to_ino(vino); | |
491 | return ilookup5(sb, t, ceph_ino_compare, &vino); | |
492 | } | |
493 | ||
494 | ||
495 | /* | |
496 | * Ceph inode. | |
497 | */ | |
10183a69 YZ |
498 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
499 | #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ | |
500 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ | |
501 | #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ | |
502 | #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ | |
503 | #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ | |
504 | #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ | |
315f2408 | 505 | #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ |
77310320 | 506 | #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ |
13c2b57d | 507 | #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ |
70220ac8 | 508 | #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ |
26544c62 | 509 | #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ |
b3f8d68f YZ |
510 | #define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */ |
511 | ||
26544c62 | 512 | |
1cf89a8d YZ |
513 | /* |
514 | * Masks of ceph inode work. | |
515 | */ | |
516 | #define CEPH_I_WORK_WRITEBACK 0 /* writeback */ | |
517 | #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */ | |
518 | #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */ | |
519 | ||
26544c62 JL |
520 | /* |
521 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode | |
522 | * and then clear it when they start succeeding. Note that we do a lockless | |
523 | * check first, and only take the lock if it looks like it needs to be changed. | |
524 | * The write submission code just takes this as a hint, so we're not too | |
525 | * worried if a few slip through in either direction. | |
526 | */ | |
527 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) | |
528 | { | |
529 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { | |
530 | spin_lock(&ci->i_ceph_lock); | |
531 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; | |
532 | spin_unlock(&ci->i_ceph_lock); | |
533 | } | |
534 | } | |
535 | ||
536 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) | |
537 | { | |
538 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { | |
539 | spin_lock(&ci->i_ceph_lock); | |
540 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; | |
541 | spin_unlock(&ci->i_ceph_lock); | |
542 | } | |
543 | } | |
3d14c5d2 | 544 | |
2f276c51 | 545 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
fdd4e158 YZ |
546 | long long release_count, |
547 | long long ordered_count) | |
de57606c | 548 | { |
74960773 AP |
549 | /* |
550 | * Makes sure operations that setup readdir cache (update page | |
551 | * cache and i_size) are strongly ordered w.r.t. the following | |
552 | * atomic64_set() operations. | |
553 | */ | |
554 | smp_mb(); | |
fdd4e158 YZ |
555 | atomic64_set(&ci->i_complete_seq[0], release_count); |
556 | atomic64_set(&ci->i_complete_seq[1], ordered_count); | |
de57606c SW |
557 | } |
558 | ||
2f276c51 | 559 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
de57606c | 560 | { |
fdd4e158 YZ |
561 | atomic64_inc(&ci->i_release_count); |
562 | } | |
563 | ||
564 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) | |
565 | { | |
566 | atomic64_inc(&ci->i_ordered_count); | |
2f276c51 | 567 | } |
de57606c | 568 | |
2f276c51 YZ |
569 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
570 | { | |
fdd4e158 YZ |
571 | return atomic64_read(&ci->i_complete_seq[0]) == |
572 | atomic64_read(&ci->i_release_count); | |
de57606c SW |
573 | } |
574 | ||
70db4f36 YZ |
575 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
576 | { | |
fdd4e158 YZ |
577 | return atomic64_read(&ci->i_complete_seq[0]) == |
578 | atomic64_read(&ci->i_release_count) && | |
579 | atomic64_read(&ci->i_complete_seq[1]) == | |
580 | atomic64_read(&ci->i_ordered_count); | |
70db4f36 YZ |
581 | } |
582 | ||
2f276c51 | 583 | static inline void ceph_dir_clear_complete(struct inode *inode) |
de57606c | 584 | { |
2f276c51 YZ |
585 | __ceph_dir_clear_complete(ceph_inode(inode)); |
586 | } | |
de57606c | 587 | |
70db4f36 | 588 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
2f276c51 | 589 | { |
fdd4e158 | 590 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
de57606c SW |
591 | } |
592 | ||
70db4f36 YZ |
593 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
594 | { | |
fdd4e158 YZ |
595 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
596 | smp_rmb(); | |
70db4f36 YZ |
597 | return ret; |
598 | } | |
de57606c SW |
599 | |
600 | /* find a specific frag @f */ | |
601 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, | |
602 | u32 f); | |
603 | ||
604 | /* | |
605 | * choose fragment for value @v. copy frag content to pfrag, if leaf | |
606 | * exists | |
607 | */ | |
608 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, | |
609 | struct ceph_inode_frag *pfrag, | |
610 | int *found); | |
611 | ||
1e9c2eb6 | 612 | static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) |
de57606c SW |
613 | { |
614 | return (struct ceph_dentry_info *)dentry->d_fsdata; | |
615 | } | |
616 | ||
de57606c SW |
617 | /* |
618 | * caps helpers | |
619 | */ | |
620 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) | |
621 | { | |
622 | return !RB_EMPTY_ROOT(&ci->i_caps); | |
623 | } | |
624 | ||
625 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); | |
626 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); | |
627 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, | |
628 | struct ceph_cap *cap); | |
629 | ||
630 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) | |
631 | { | |
632 | int issued; | |
be655596 | 633 | spin_lock(&ci->i_ceph_lock); |
de57606c | 634 | issued = __ceph_caps_issued(ci, NULL); |
be655596 | 635 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
636 | return issued; |
637 | } | |
638 | ||
639 | static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, | |
640 | int touch) | |
641 | { | |
642 | int r; | |
be655596 | 643 | spin_lock(&ci->i_ceph_lock); |
de57606c | 644 | r = __ceph_caps_issued_mask(ci, mask, touch); |
be655596 | 645 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
646 | return r; |
647 | } | |
648 | ||
649 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) | |
650 | { | |
651 | return ci->i_dirty_caps | ci->i_flushing_caps; | |
652 | } | |
f66fd9f0 YZ |
653 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
654 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); | |
655 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, | |
656 | struct ceph_cap_flush **pcf); | |
de57606c | 657 | |
9563f88c YZ |
658 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
659 | struct ceph_cap *ocap, int mask); | |
de57606c SW |
660 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
661 | extern int __ceph_caps_used(struct ceph_inode_info *ci); | |
662 | ||
663 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); | |
664 | ||
665 | /* | |
666 | * wanted, by virtue of open file modes AND cap refs (buffered/cached data) | |
667 | */ | |
668 | static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) | |
669 | { | |
670 | int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); | |
671 | if (w & CEPH_CAP_FILE_BUFFER) | |
672 | w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ | |
673 | return w; | |
674 | } | |
675 | ||
676 | /* what the mds thinks we want */ | |
c1944fed | 677 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
de57606c | 678 | |
37151668 YS |
679 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
680 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); | |
fe33032d YZ |
681 | extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, |
682 | struct ceph_mount_options *fsopt); | |
e30ee581 | 683 | extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 684 | struct ceph_cap_reservation *ctx, int need); |
7bf8f736 | 685 | extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 686 | struct ceph_cap_reservation *ctx); |
3d14c5d2 | 687 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
de57606c | 688 | int *total, int *avail, int *used, |
85ccce43 | 689 | int *reserved, int *min); |
de57606c | 690 | |
de57606c | 691 | |
de57606c SW |
692 | |
693 | /* | |
694 | * we keep buffered readdir results attached to file->private_data | |
695 | */ | |
4918b6d1 | 696 | #define CEPH_F_SYNC 1 |
9cfa1098 | 697 | #define CEPH_F_ATEND 2 |
4918b6d1 | 698 | |
de57606c | 699 | struct ceph_file_info { |
252c6728 SW |
700 | short fmode; /* initialized on open */ |
701 | short flags; /* CEPH_F_* */ | |
de57606c | 702 | |
5d988308 YZ |
703 | spinlock_t rw_contexts_lock; |
704 | struct list_head rw_contexts; | |
bb48bd4d CX |
705 | }; |
706 | ||
707 | struct ceph_dir_file_info { | |
708 | struct ceph_file_info file_info; | |
5d988308 | 709 | |
de57606c SW |
710 | /* readdir: position within the dir */ |
711 | u32 frag; | |
712 | struct ceph_mds_request *last_readdir; | |
de57606c SW |
713 | |
714 | /* readdir: position within a frag */ | |
f0494206 | 715 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
de57606c | 716 | char *last_name; /* last entry in previous chunk */ |
fdd4e158 YZ |
717 | long long dir_release_count; |
718 | long long dir_ordered_count; | |
719 | int readdir_cache_idx; | |
de57606c SW |
720 | |
721 | /* used for -o dirstat read() on directory thing */ | |
722 | char *dir_info; | |
723 | int dir_info_len; | |
724 | }; | |
725 | ||
5d988308 YZ |
726 | struct ceph_rw_context { |
727 | struct list_head list; | |
728 | struct task_struct *thread; | |
729 | int caps; | |
730 | }; | |
731 | ||
732 | #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ | |
733 | struct ceph_rw_context _name = { \ | |
734 | .thread = current, \ | |
735 | .caps = _caps, \ | |
736 | } | |
737 | ||
738 | static inline void ceph_add_rw_context(struct ceph_file_info *cf, | |
739 | struct ceph_rw_context *ctx) | |
740 | { | |
741 | spin_lock(&cf->rw_contexts_lock); | |
742 | list_add(&ctx->list, &cf->rw_contexts); | |
743 | spin_unlock(&cf->rw_contexts_lock); | |
744 | } | |
745 | ||
746 | static inline void ceph_del_rw_context(struct ceph_file_info *cf, | |
747 | struct ceph_rw_context *ctx) | |
748 | { | |
749 | spin_lock(&cf->rw_contexts_lock); | |
750 | list_del(&ctx->list); | |
751 | spin_unlock(&cf->rw_contexts_lock); | |
752 | } | |
753 | ||
754 | static inline struct ceph_rw_context* | |
755 | ceph_find_rw_context(struct ceph_file_info *cf) | |
756 | { | |
757 | struct ceph_rw_context *ctx, *found = NULL; | |
758 | spin_lock(&cf->rw_contexts_lock); | |
759 | list_for_each_entry(ctx, &cf->rw_contexts, list) { | |
760 | if (ctx->thread == current) { | |
761 | found = ctx; | |
762 | break; | |
763 | } | |
764 | } | |
765 | spin_unlock(&cf->rw_contexts_lock); | |
766 | return found; | |
767 | } | |
768 | ||
fdd4e158 YZ |
769 | struct ceph_readdir_cache_control { |
770 | struct page *page; | |
771 | struct dentry **dentries; | |
772 | int index; | |
773 | }; | |
de57606c | 774 | |
de57606c SW |
775 | /* |
776 | * A "snap realm" describes a subset of the file hierarchy sharing | |
777 | * the same set of snapshots that apply to it. The realms themselves | |
778 | * are organized into a hierarchy, such that children inherit (some of) | |
779 | * the snapshots of their parents. | |
780 | * | |
781 | * All inodes within the realm that have capabilities are linked into a | |
782 | * per-realm list. | |
783 | */ | |
784 | struct ceph_snap_realm { | |
785 | u64 ino; | |
e3161f17 | 786 | struct inode *inode; |
de57606c | 787 | atomic_t nref; |
a105f00c SW |
788 | struct rb_node node; |
789 | ||
de57606c SW |
790 | u64 created, seq; |
791 | u64 parent_ino; | |
792 | u64 parent_since; /* snapid when our current parent became so */ | |
793 | ||
794 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ | |
aa711ee3 | 795 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
de57606c | 796 | u64 *snaps; /* snaps specific to this realm */ |
aa711ee3 | 797 | u32 num_snaps; |
de57606c SW |
798 | |
799 | struct ceph_snap_realm *parent; | |
800 | struct list_head children; /* list of child realms */ | |
801 | struct list_head child_item; | |
802 | ||
803 | struct list_head empty_item; /* if i have ref==0 */ | |
804 | ||
ae00d4f3 SW |
805 | struct list_head dirty_item; /* if realm needs new context */ |
806 | ||
de57606c SW |
807 | /* the current set of snaps for this realm */ |
808 | struct ceph_snap_context *cached_context; | |
809 | ||
810 | struct list_head inodes_with_caps; | |
811 | spinlock_t inodes_with_caps_lock; | |
812 | }; | |
813 | ||
3d14c5d2 | 814 | static inline int default_congestion_kb(void) |
de57606c | 815 | { |
3d14c5d2 YS |
816 | int congestion_kb; |
817 | ||
818 | /* | |
819 | * Copied from NFS | |
820 | * | |
821 | * congestion size, scale with available memory. | |
822 | * | |
823 | * 64MB: 8192k | |
824 | * 128MB: 11585k | |
825 | * 256MB: 16384k | |
826 | * 512MB: 23170k | |
827 | * 1GB: 32768k | |
828 | * 2GB: 46340k | |
829 | * 4GB: 65536k | |
830 | * 8GB: 92681k | |
831 | * 16GB: 131072k | |
832 | * | |
833 | * This allows larger machines to have larger/more transfers. | |
834 | * Limit the default to 256M | |
835 | */ | |
ca79b0c2 | 836 | congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); |
3d14c5d2 YS |
837 | if (congestion_kb > 256*1024) |
838 | congestion_kb = 256*1024; | |
839 | ||
840 | return congestion_kb; | |
de57606c SW |
841 | } |
842 | ||
843 | ||
844 | ||
845 | /* snap.c */ | |
846 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, | |
847 | u64 ino); | |
848 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, | |
849 | struct ceph_snap_realm *realm); | |
850 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, | |
851 | struct ceph_snap_realm *realm); | |
852 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, | |
982d6011 YZ |
853 | void *p, void *e, bool deletion, |
854 | struct ceph_snap_realm **realm_ret); | |
de57606c | 855 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
2600d2dd | 856 | struct ceph_mds_session *session, |
de57606c | 857 | struct ceph_msg *msg); |
fc837c8f | 858 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
de57606c SW |
859 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
860 | struct ceph_cap_snap *capsnap); | |
861 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); | |
862 | ||
75c9627e YZ |
863 | extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, |
864 | u64 snap); | |
865 | extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, | |
866 | struct ceph_snapid_map *sm); | |
867 | extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); | |
868 | extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); | |
869 | ||
870 | ||
de57606c SW |
871 | /* |
872 | * a cap_snap is "pending" if it is still awaiting an in-progress | |
873 | * sync write (that may/may not still update size, mtime, etc.). | |
874 | */ | |
875 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) | |
876 | { | |
877 | return !list_empty(&ci->i_cap_snaps) && | |
86056090 YZ |
878 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
879 | ci_item)->writing; | |
de57606c SW |
880 | } |
881 | ||
de57606c SW |
882 | /* inode.c */ |
883 | extern const struct inode_operations ceph_file_iops; | |
884 | ||
885 | extern struct inode *ceph_alloc_inode(struct super_block *sb); | |
886 | extern void ceph_destroy_inode(struct inode *inode); | |
cfa6d412 | 887 | extern void ceph_free_inode(struct inode *inode); |
9f12bd11 | 888 | extern int ceph_drop_inode(struct inode *inode); |
de57606c SW |
889 | |
890 | extern struct inode *ceph_get_inode(struct super_block *sb, | |
891 | struct ceph_vino vino); | |
892 | extern struct inode *ceph_get_snapdir(struct inode *parent); | |
893 | extern int ceph_fill_file_size(struct inode *inode, int issued, | |
894 | u32 truncate_seq, u64 truncate_size, u64 size); | |
895 | extern void ceph_fill_file_time(struct inode *inode, int issued, | |
9bbeab41 AB |
896 | u64 time_warp_seq, struct timespec64 *ctime, |
897 | struct timespec64 *mtime, | |
898 | struct timespec64 *atime); | |
de57606c | 899 | extern int ceph_fill_trace(struct super_block *sb, |
f5a03b08 | 900 | struct ceph_mds_request *req); |
de57606c SW |
901 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
902 | struct ceph_mds_session *session); | |
903 | ||
904 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); | |
905 | ||
efb0ca76 | 906 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
b415bf4f | 907 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
3c6f6b79 | 908 | extern void ceph_queue_vmtruncate(struct inode *inode); |
3c6f6b79 SW |
909 | extern void ceph_queue_invalidate(struct inode *inode); |
910 | extern void ceph_queue_writeback(struct inode *inode); | |
3e1d0452 | 911 | extern void ceph_async_iput(struct inode *inode); |
de57606c | 912 | |
01deead0 YZ |
913 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
914 | int mask, bool force); | |
915 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) | |
916 | { | |
917 | return __ceph_do_getattr(inode, NULL, mask, force); | |
918 | } | |
10556cb2 | 919 | extern int ceph_permission(struct inode *inode, int mask); |
a26fecca | 920 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
de57606c | 921 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
a528d35e DH |
922 | extern int ceph_getattr(const struct path *path, struct kstat *stat, |
923 | u32 request_mask, unsigned int flags); | |
de57606c SW |
924 | |
925 | /* xattr.c */ | |
a26fecca | 926 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
7221fe4c | 927 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
de57606c | 928 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
de57606c SW |
929 | extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
930 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); | |
b1ee94aa | 931 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
de57606c | 932 | |
5c31e92d YZ |
933 | struct ceph_acl_sec_ctx { |
934 | #ifdef CONFIG_CEPH_FS_POSIX_ACL | |
935 | void *default_acl; | |
936 | void *acl; | |
ac6713cc YZ |
937 | #endif |
938 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL | |
939 | void *sec_ctx; | |
940 | u32 sec_ctxlen; | |
5c31e92d YZ |
941 | #endif |
942 | struct ceph_pagelist *pagelist; | |
943 | }; | |
944 | ||
315f2408 YZ |
945 | #ifdef CONFIG_SECURITY |
946 | extern bool ceph_security_xattr_deadlock(struct inode *in); | |
947 | extern bool ceph_security_xattr_wanted(struct inode *in); | |
948 | #else | |
949 | static inline bool ceph_security_xattr_deadlock(struct inode *in) | |
950 | { | |
951 | return false; | |
952 | } | |
953 | static inline bool ceph_security_xattr_wanted(struct inode *in) | |
954 | { | |
955 | return false; | |
956 | } | |
957 | #endif | |
958 | ||
ac6713cc YZ |
959 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL |
960 | extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
961 | struct ceph_acl_sec_ctx *ctx); | |
962 | extern void ceph_security_invalidate_secctx(struct inode *inode); | |
963 | #else | |
964 | static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
965 | struct ceph_acl_sec_ctx *ctx) | |
966 | { | |
967 | return 0; | |
968 | } | |
969 | static inline void ceph_security_invalidate_secctx(struct inode *inode) | |
970 | { | |
971 | } | |
972 | #endif | |
973 | ||
5c31e92d | 974 | void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); |
7221fe4c | 975 | |
5c31e92d | 976 | /* acl.c */ |
7221fe4c GZ |
977 | #ifdef CONFIG_CEPH_FS_POSIX_ACL |
978 | ||
979 | struct posix_acl *ceph_get_acl(struct inode *, int); | |
72466d0b | 980 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
b1ee94aa | 981 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d YZ |
982 | struct ceph_acl_sec_ctx *as_ctx); |
983 | void ceph_init_inode_acls(struct inode *inode, | |
984 | struct ceph_acl_sec_ctx *as_ctx); | |
c969d9bf GZ |
985 | |
986 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
987 | { | |
988 | forget_all_cached_acls(inode); | |
989 | } | |
7221fe4c GZ |
990 | |
991 | #else | |
992 | ||
993 | #define ceph_get_acl NULL | |
72466d0b | 994 | #define ceph_set_acl NULL |
7221fe4c | 995 | |
b1ee94aa | 996 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d | 997 | struct ceph_acl_sec_ctx *as_ctx) |
7221fe4c GZ |
998 | { |
999 | return 0; | |
1000 | } | |
b1ee94aa | 1001 | static inline void ceph_init_inode_acls(struct inode *inode, |
5c31e92d | 1002 | struct ceph_acl_sec_ctx *as_ctx) |
b1ee94aa YZ |
1003 | { |
1004 | } | |
7221fe4c GZ |
1005 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
1006 | { | |
1007 | return 0; | |
1008 | } | |
1009 | ||
1010 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
1011 | { | |
1012 | } | |
1013 | ||
1014 | #endif | |
1015 | ||
de57606c SW |
1016 | /* caps.c */ |
1017 | extern const char *ceph_cap_string(int c); | |
1018 | extern void ceph_handle_caps(struct ceph_mds_session *session, | |
1019 | struct ceph_msg *msg); | |
d9df2783 YZ |
1020 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
1021 | struct ceph_cap_reservation *ctx); | |
1022 | extern void ceph_add_cap(struct inode *inode, | |
1023 | struct ceph_mds_session *session, u64 cap_id, | |
1024 | int fmode, unsigned issued, unsigned wanted, | |
1025 | unsigned cap, unsigned seq, u64 realmino, int flags, | |
1026 | struct ceph_cap **new_cap); | |
a096b09a | 1027 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
d6e47819 | 1028 | extern void __ceph_remove_caps(struct ceph_inode_info *ci); |
37151668 YS |
1029 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
1030 | struct ceph_cap *cap); | |
9215aeea | 1031 | extern int ceph_is_any_caps(struct inode *inode); |
de57606c | 1032 | |
f1a3d572 | 1033 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
02c24a82 JB |
1034 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
1035 | int datasync); | |
e548e9b9 YZ |
1036 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1037 | struct ceph_mds_session *session); | |
de57606c SW |
1038 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1039 | struct ceph_mds_session *session); | |
2bc50259 GF |
1040 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
1041 | int mds); | |
de57606c SW |
1042 | extern int ceph_get_cap_mds(struct inode *inode); |
1043 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); | |
1044 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); | |
1045 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, | |
1046 | struct ceph_snap_context *snapc); | |
ed9b430c YZ |
1047 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
1048 | struct ceph_mds_session **psession); | |
efb0ca76 | 1049 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
de57606c SW |
1050 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
1051 | struct ceph_mds_session *session); | |
afcdaea3 SW |
1052 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
1053 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); | |
6ef0bc6d | 1054 | extern int ceph_drop_caps_for_unlink(struct inode *inode); |
de57606c SW |
1055 | extern int ceph_encode_inode_release(void **p, struct inode *inode, |
1056 | int mds, int drop, int unless, int force); | |
1057 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, | |
ca6c8ae0 | 1058 | struct inode *dir, |
de57606c SW |
1059 | int mds, int drop, int unless); |
1060 | ||
1061 | extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, | |
31c542a1 | 1062 | loff_t endoff, int *got, struct page **pinned_page); |
2b1ac852 | 1063 | extern int ceph_try_get_caps(struct ceph_inode_info *ci, |
2ee9dd95 | 1064 | int need, int want, bool nonblock, int *got); |
de57606c SW |
1065 | |
1066 | /* for counting open files by mode */ | |
774a6a11 | 1067 | extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); |
de57606c SW |
1068 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); |
1069 | ||
1070 | /* addr.c */ | |
1071 | extern const struct address_space_operations ceph_aops; | |
1072 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); | |
10183a69 YZ |
1073 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
1074 | extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); | |
1075 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); | |
de57606c SW |
1076 | |
1077 | /* file.c */ | |
1078 | extern const struct file_operations ceph_file_fops; | |
9e0eb85d | 1079 | |
77310320 | 1080 | extern int ceph_renew_caps(struct inode *inode); |
de57606c | 1081 | extern int ceph_open(struct inode *inode, struct file *file); |
5ef50c3b | 1082 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
44907d79 | 1083 | struct file *file, unsigned flags, umode_t mode); |
de57606c | 1084 | extern int ceph_release(struct inode *inode, struct file *filp); |
31c542a1 YZ |
1085 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
1086 | char *data, size_t len); | |
55f2a045 | 1087 | |
de57606c SW |
1088 | /* dir.c */ |
1089 | extern const struct file_operations ceph_dir_fops; | |
38c48b5f | 1090 | extern const struct file_operations ceph_snapdir_fops; |
de57606c | 1091 | extern const struct inode_operations ceph_dir_iops; |
38c48b5f | 1092 | extern const struct inode_operations ceph_snapdir_iops; |
18fc8abd | 1093 | extern const struct dentry_operations ceph_dentry_ops; |
de57606c | 1094 | |
f3c4ebe6 | 1095 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
de57606c | 1096 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
468640e3 SW |
1097 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
1098 | struct dentry *dentry, int err); | |
de57606c SW |
1099 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
1100 | struct dentry *dentry, int err); | |
1101 | ||
37c4efc1 YZ |
1102 | extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); |
1103 | extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); | |
81a6cf2d | 1104 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
37c4efc1 | 1105 | extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); |
e5f86dc3 | 1106 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
fdd4e158 | 1107 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
de57606c | 1108 | |
de57606c SW |
1109 | /* ioctl.c */ |
1110 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); | |
1111 | ||
1112 | /* export.c */ | |
1113 | extern const struct export_operations ceph_export_ops; | |
3886274a | 1114 | struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); |
de57606c | 1115 | |
40819f6f | 1116 | /* locks.c */ |
eb13e832 | 1117 | extern __init void ceph_flock_init(void); |
40819f6f GF |
1118 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
1119 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); | |
1120 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); | |
39be95e9 JS |
1121 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
1122 | struct ceph_filelock *flocks, | |
1123 | int num_fcntl_locks, | |
1124 | int num_flock_locks); | |
1125 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, | |
1126 | struct ceph_pagelist *pagelist, | |
1127 | int num_fcntl_locks, int num_flock_locks); | |
40819f6f | 1128 | |
3d14c5d2 YS |
1129 | /* debugfs.c */ |
1130 | extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); | |
1131 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); | |
1132 | ||
fb18a575 | 1133 | /* quota.c */ |
d557c48d LH |
1134 | static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) |
1135 | { | |
1136 | return ci->i_max_files || ci->i_max_bytes; | |
1137 | } | |
1138 | ||
1139 | extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); | |
1140 | ||
1141 | static inline void __ceph_update_quota(struct ceph_inode_info *ci, | |
1142 | u64 max_bytes, u64 max_files) | |
1143 | { | |
1144 | bool had_quota, has_quota; | |
1145 | had_quota = __ceph_has_any_quota(ci); | |
1146 | ci->i_max_bytes = max_bytes; | |
1147 | ci->i_max_files = max_files; | |
1148 | has_quota = __ceph_has_any_quota(ci); | |
1149 | ||
1150 | if (had_quota != has_quota) | |
1151 | ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); | |
1152 | } | |
1153 | ||
fb18a575 LH |
1154 | extern void ceph_handle_quota(struct ceph_mds_client *mdsc, |
1155 | struct ceph_mds_session *session, | |
1156 | struct ceph_msg *msg); | |
b7a29217 | 1157 | extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); |
cafe21a4 | 1158 | extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); |
2b83845f LH |
1159 | extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, |
1160 | loff_t newlen); | |
1ab302a0 LH |
1161 | extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, |
1162 | loff_t newlen); | |
9122eed5 LH |
1163 | extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, |
1164 | struct kstatfs *buf); | |
0c44a8e0 | 1165 | extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); |
fb18a575 | 1166 | |
de57606c | 1167 | #endif /* _FS_CEPH_SUPER_H */ |