Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
de57606c SW |
2 | #ifndef _FS_CEPH_SUPER_H |
3 | #define _FS_CEPH_SUPER_H | |
4 | ||
3d14c5d2 | 5 | #include <linux/ceph/ceph_debug.h> |
de57606c SW |
6 | |
7 | #include <asm/unaligned.h> | |
8 | #include <linux/backing-dev.h> | |
9 | #include <linux/completion.h> | |
10 | #include <linux/exportfs.h> | |
11 | #include <linux/fs.h> | |
12 | #include <linux/mempool.h> | |
13 | #include <linux/pagemap.h> | |
14 | #include <linux/wait.h> | |
f1a3d572 | 15 | #include <linux/writeback.h> |
5a0e3ad6 | 16 | #include <linux/slab.h> |
c969d9bf | 17 | #include <linux/posix_acl.h> |
805692d0 | 18 | #include <linux/refcount.h> |
668959a5 | 19 | #include <linux/security.h> |
de57606c | 20 | |
3d14c5d2 | 21 | #include <linux/ceph/libceph.h> |
de57606c | 22 | |
99ccbd22 MT |
23 | #ifdef CONFIG_CEPH_FSCACHE |
24 | #include <linux/fscache.h> | |
25 | #endif | |
26 | ||
de57606c SW |
27 | /* f_type in struct statfs */ |
28 | #define CEPH_SUPER_MAGIC 0x00c36400 | |
29 | ||
30 | /* large granularity for statfs utilization stats to facilitate | |
31 | * large volume sizes on 32-bit machines. */ | |
92a49fb0 | 32 | #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ |
de57606c SW |
33 | #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) |
34 | ||
131d7eb4 | 35 | #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blacklisted */ |
3d14c5d2 YS |
36 | #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ |
37 | #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ | |
38 | #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ | |
ad1fee96 | 39 | #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ |
a40dc6cc | 40 | #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ |
99ccbd22 | 41 | #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ |
10183a69 | 42 | #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ |
e9e427f0 | 43 | #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ |
9122eed5 | 44 | #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ |
ea4cdc54 | 45 | #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ |
2ccb4546 | 46 | #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ |
6a259382 | 47 | |
6f9718fe LH |
48 | #define CEPH_MOUNT_OPT_DEFAULT \ |
49 | (CEPH_MOUNT_OPT_DCACHE | \ | |
50 | CEPH_MOUNT_OPT_NOCOPYFROM) | |
de57606c | 51 | |
3d14c5d2 | 52 | #define ceph_set_mount_opt(fsc, opt) \ |
2ccb4546 JL |
53 | (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt |
54 | #define ceph_clear_mount_opt(fsc, opt) \ | |
55 | (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt | |
3d14c5d2 YS |
56 | #define ceph_test_mount_opt(fsc, opt) \ |
57 | (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) | |
de57606c | 58 | |
aa187926 YZ |
59 | /* max size of osd read request, limited by libceph */ |
60 | #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN | |
95cca2b4 YZ |
61 | /* osd has a configurable limitaion of max write size. |
62 | * CEPH_MSG_MAX_DATA_LEN should be small enough. */ | |
63 | #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN | |
7c94ba27 | 64 | #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ |
3d14c5d2 YS |
65 | #define CEPH_MAX_READDIR_DEFAULT 1024 |
66 | #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) | |
67 | #define CEPH_SNAPDIRNAME_DEFAULT ".snap" | |
de57606c | 68 | |
4214fb15 YZ |
69 | /* |
70 | * Delay telling the MDS we no longer want caps, in case we reopen | |
71 | * the file. Delay a minimum amount of time, even if we send a cap | |
72 | * message for some other reason. Otherwise, take the oppotunity to | |
73 | * update the mds to avoid sending another message later. | |
74 | */ | |
75 | #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ | |
76 | #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ | |
77 | ||
3d14c5d2 | 78 | struct ceph_mount_options { |
ad8c28a9 | 79 | unsigned int flags; |
3d14c5d2 | 80 | |
ad8c28a9 JL |
81 | unsigned int wsize; /* max write size */ |
82 | unsigned int rsize; /* max read size */ | |
83 | unsigned int rasize; /* max readahead */ | |
84 | unsigned int congestion_kb; /* max writeback in flight */ | |
85 | unsigned int caps_wanted_delay_min, caps_wanted_delay_max; | |
fe33032d | 86 | int caps_max; |
ad8c28a9 JL |
87 | unsigned int max_readdir; /* max readdir result (entries) */ |
88 | unsigned int max_readdir_bytes; /* max readdir result (bytes) */ | |
de57606c | 89 | |
3d14c5d2 YS |
90 | /* |
91 | * everything above this point can be memcmp'd; everything below | |
92 | * is handled in compare_mount_options() | |
93 | */ | |
de57606c | 94 | |
3d14c5d2 | 95 | char *snapdir_name; /* default ".snap" */ |
430afbad | 96 | char *mds_namespace; /* default NULL */ |
b27a939e | 97 | char *server_path; /* default NULL (means "/") */ |
1d8f8360 | 98 | char *fscache_uniq; /* default NULL */ |
3d14c5d2 | 99 | }; |
de57606c | 100 | |
3d14c5d2 | 101 | struct ceph_fs_client { |
de57606c SW |
102 | struct super_block *sb; |
103 | ||
3d14c5d2 YS |
104 | struct ceph_mount_options *mount_options; |
105 | struct ceph_client *client; | |
de57606c | 106 | |
3d14c5d2 | 107 | unsigned long mount_state; |
81f148a9 | 108 | |
131d7eb4 YZ |
109 | unsigned long last_auto_reconnect; |
110 | bool blacklisted; | |
111 | ||
78beb0ff LH |
112 | bool have_copy_from2; |
113 | ||
81f148a9 | 114 | u32 filp_gen; |
719784ba | 115 | loff_t max_file_size; |
85ccce43 | 116 | |
3d14c5d2 | 117 | struct ceph_mds_client *mdsc; |
de57606c SW |
118 | |
119 | /* writeback */ | |
120 | mempool_t *wb_pagevec_pool; | |
e3ec8d68 YZ |
121 | atomic_long_t writeback_count; |
122 | ||
1cf89a8d | 123 | struct workqueue_struct *inode_wq; |
e3ec8d68 | 124 | struct workqueue_struct *cap_wq; |
de57606c | 125 | |
0743304d | 126 | #ifdef CONFIG_DEBUG_FS |
3d14c5d2 | 127 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
2baba250 | 128 | struct dentry *debugfs_congestion_kb; |
06edf046 | 129 | struct dentry *debugfs_bdi; |
3d14c5d2 | 130 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
14ed9703 | 131 | struct dentry *debugfs_mds_sessions; |
0743304d | 132 | #endif |
99ccbd22 MT |
133 | |
134 | #ifdef CONFIG_CEPH_FSCACHE | |
135 | struct fscache_cookie *fscache; | |
99ccbd22 | 136 | #endif |
de57606c SW |
137 | }; |
138 | ||
3d14c5d2 | 139 | |
de57606c SW |
140 | /* |
141 | * File i/o capability. This tracks shared state with the metadata | |
142 | * server that allows us to cache or writeback attributes or to read | |
143 | * and write data. For any given inode, we should have one or more | |
144 | * capabilities, one issued by each metadata server, and our | |
145 | * cumulative access is the OR of all issued capabilities. | |
146 | * | |
147 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds | |
148 | * session capability lists. | |
149 | */ | |
150 | struct ceph_cap { | |
151 | struct ceph_inode_info *ci; | |
152 | struct rb_node ci_node; /* per-ci cap tree */ | |
153 | struct ceph_mds_session *session; | |
154 | struct list_head session_caps; /* per-session caplist */ | |
de57606c | 155 | u64 cap_id; /* unique cap id (mds provided) */ |
745a8e3b YZ |
156 | union { |
157 | /* in-use caps */ | |
158 | struct { | |
159 | int issued; /* latest, from the mds */ | |
160 | int implemented; /* implemented superset of | |
161 | issued (for revocation) */ | |
162 | int mds, mds_wanted; | |
163 | }; | |
164 | /* caps to release */ | |
165 | struct { | |
166 | u64 cap_ino; | |
167 | int queue_release; | |
168 | }; | |
169 | }; | |
685f9a5d SW |
170 | u32 seq, issue_seq, mseq; |
171 | u32 cap_gen; /* active/stale cycle */ | |
de57606c SW |
172 | unsigned long last_used; |
173 | struct list_head caps_item; | |
174 | }; | |
175 | ||
a0d93e32 YZ |
176 | #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */ |
177 | #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */ | |
178 | #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */ | |
de57606c | 179 | |
0e294387 YZ |
180 | struct ceph_cap_flush { |
181 | u64 tid; | |
182 | int caps; /* 0 means capsnap */ | |
c8799fc4 | 183 | bool wake; /* wake up flush waiters when finish ? */ |
0e294387 YZ |
184 | struct list_head g_list; // global |
185 | struct list_head i_list; // per inode | |
186 | }; | |
187 | ||
de57606c SW |
188 | /* |
189 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, | |
190 | * we first complete any in-process sync writes and writeback any dirty | |
191 | * data before flushing the snapped state (tracked here) back to the MDS. | |
192 | */ | |
193 | struct ceph_cap_snap { | |
805692d0 | 194 | refcount_t nref; |
0e294387 | 195 | struct list_head ci_item; |
de57606c | 196 | |
0e294387 YZ |
197 | struct ceph_cap_flush cap_flush; |
198 | ||
199 | u64 follows; | |
de57606c SW |
200 | int issued, dirty; |
201 | struct ceph_snap_context *context; | |
202 | ||
5706b27d | 203 | umode_t mode; |
05cb11c1 EB |
204 | kuid_t uid; |
205 | kgid_t gid; | |
de57606c | 206 | |
4a625be4 | 207 | struct ceph_buffer *xattr_blob; |
de57606c SW |
208 | u64 xattr_version; |
209 | ||
210 | u64 size; | |
176c77c9 | 211 | u64 change_attr; |
ec62b894 | 212 | struct timespec64 mtime, atime, ctime, btime; |
de57606c | 213 | u64 time_warp_seq; |
5f743e45 YZ |
214 | u64 truncate_size; |
215 | u32 truncate_seq; | |
de57606c SW |
216 | int writing; /* a sync write is still in progress */ |
217 | int dirty_pages; /* dirty pages awaiting writeback */ | |
e20d258d | 218 | bool inline_data; |
86056090 | 219 | bool need_flush; |
de57606c SW |
220 | }; |
221 | ||
222 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) | |
223 | { | |
805692d0 | 224 | if (refcount_dec_and_test(&capsnap->nref)) { |
4a625be4 SW |
225 | if (capsnap->xattr_blob) |
226 | ceph_buffer_put(capsnap->xattr_blob); | |
de57606c | 227 | kfree(capsnap); |
4a625be4 | 228 | } |
de57606c SW |
229 | } |
230 | ||
231 | /* | |
232 | * The frag tree describes how a directory is fragmented, potentially across | |
233 | * multiple metadata servers. It is also used to indicate points where | |
234 | * metadata authority is delegated, and whether/where metadata is replicated. | |
235 | * | |
236 | * A _leaf_ frag will be present in the i_fragtree IFF there is | |
237 | * delegation info. That is, if mds >= 0 || ndist > 0. | |
238 | */ | |
239 | #define CEPH_MAX_DIRFRAG_REP 4 | |
240 | ||
241 | struct ceph_inode_frag { | |
242 | struct rb_node node; | |
243 | ||
244 | /* fragtree state */ | |
245 | u32 frag; | |
246 | int split_by; /* i.e. 2^(split_by) children */ | |
247 | ||
248 | /* delegation and replication info */ | |
249 | int mds; /* -1 if same authority as parent */ | |
250 | int ndist; /* >0 if replicated */ | |
251 | int dist[CEPH_MAX_DIRFRAG_REP]; | |
252 | }; | |
253 | ||
254 | /* | |
255 | * We cache inode xattrs as an encoded blob until they are first used, | |
256 | * at which point we parse them into an rbtree. | |
257 | */ | |
258 | struct ceph_inode_xattr { | |
259 | struct rb_node node; | |
260 | ||
261 | const char *name; | |
262 | int name_len; | |
263 | const char *val; | |
264 | int val_len; | |
265 | int dirty; | |
266 | ||
267 | int should_free_name; | |
268 | int should_free_val; | |
269 | }; | |
270 | ||
3d14c5d2 YS |
271 | /* |
272 | * Ceph dentry state | |
273 | */ | |
274 | struct ceph_dentry_info { | |
37c4efc1 | 275 | struct dentry *dentry; |
3d14c5d2 | 276 | struct ceph_mds_session *lease_session; |
37c4efc1 YZ |
277 | struct list_head lease_list; |
278 | unsigned flags; | |
97aeb6bf YZ |
279 | int lease_shared_gen; |
280 | u32 lease_gen; | |
3d14c5d2 YS |
281 | u32 lease_seq; |
282 | unsigned long lease_renew_after, lease_renew_from; | |
9b16f03c | 283 | unsigned long time; |
3d14c5d2 YS |
284 | u64 offset; |
285 | }; | |
286 | ||
37c4efc1 YZ |
287 | #define CEPH_DENTRY_REFERENCED 1 |
288 | #define CEPH_DENTRY_LEASE_LIST 2 | |
289 | #define CEPH_DENTRY_SHRINK_LIST 4 | |
f5e17aed | 290 | #define CEPH_DENTRY_PRIMARY_LINK 8 |
37c4efc1 | 291 | |
de57606c SW |
292 | struct ceph_inode_xattrs_info { |
293 | /* | |
294 | * (still encoded) xattr blob. we avoid the overhead of parsing | |
295 | * this until someone actually calls getxattr, etc. | |
296 | * | |
297 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == | |
298 | * NULL means we don't know. | |
299 | */ | |
300 | struct ceph_buffer *blob, *prealloc_blob; | |
301 | ||
302 | struct rb_root index; | |
303 | bool dirty; | |
304 | int count; | |
305 | int names_size; | |
306 | int vals_size; | |
307 | u64 version, index_version; | |
308 | }; | |
309 | ||
310 | /* | |
311 | * Ceph inode. | |
312 | */ | |
de57606c SW |
313 | struct ceph_inode_info { |
314 | struct ceph_vino i_vino; /* ceph ino + snap */ | |
315 | ||
be655596 SW |
316 | spinlock_t i_ceph_lock; |
317 | ||
de57606c | 318 | u64 i_version; |
31c542a1 | 319 | u64 i_inline_version; |
de57606c SW |
320 | u32 i_time_warp_seq; |
321 | ||
891f3f5a | 322 | unsigned long i_ceph_flags; |
fdd4e158 YZ |
323 | atomic64_t i_release_count; |
324 | atomic64_t i_ordered_count; | |
325 | atomic64_t i_complete_seq[2]; | |
de57606c | 326 | |
6c0f3af7 | 327 | struct ceph_dir_layout i_dir_layout; |
de57606c | 328 | struct ceph_file_layout i_layout; |
785892fe | 329 | struct ceph_file_layout i_cached_layout; // for async creates |
de57606c SW |
330 | char *i_symlink; |
331 | ||
332 | /* for dirs */ | |
9bbeab41 | 333 | struct timespec64 i_rctime; |
de57606c SW |
334 | u64 i_rbytes, i_rfiles, i_rsubdirs; |
335 | u64 i_files, i_subdirs; | |
de57606c | 336 | |
fb18a575 LH |
337 | /* quotas */ |
338 | u64 i_max_bytes, i_max_files; | |
339 | ||
08796873 YZ |
340 | s32 i_dir_pin; |
341 | ||
de57606c | 342 | struct rb_root i_fragtree; |
1b1bc16d | 343 | int i_fragtree_nsplits; |
de57606c SW |
344 | struct mutex i_fragtree_mutex; |
345 | ||
346 | struct ceph_inode_xattrs_info i_xattrs; | |
347 | ||
be655596 | 348 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
de57606c SW |
349 | * s_mutex. */ |
350 | struct rb_root i_caps; /* cap list */ | |
351 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ | |
352 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ | |
353 | struct list_head i_dirty_item, i_flushing_item; | |
de57606c SW |
354 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
355 | * overlapping, pipelined cap flushes to the mds. we can probably | |
356 | * reduce the tid to 8 bits if we're concerned about inode size. */ | |
f66fd9f0 | 357 | struct ceph_cap_flush *i_prealloc_cap_flush; |
e4500b5e | 358 | struct list_head i_cap_flush_list; |
de57606c | 359 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
de57606c SW |
360 | unsigned long i_hold_caps_max; /* jiffies */ |
361 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ | |
de57606c SW |
362 | struct ceph_cap_reservation i_cap_migration_resv; |
363 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ | |
7d8cb26d SW |
364 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
365 | dirty|flushing caps */ | |
de57606c SW |
366 | unsigned i_snap_caps; /* cap bits for snapped files */ |
367 | ||
719a2514 YZ |
368 | unsigned long i_last_rd; |
369 | unsigned long i_last_wr; | |
774a6a11 | 370 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
de57606c | 371 | |
b0d7c223 | 372 | struct mutex i_truncate_mutex; |
de57606c SW |
373 | u32 i_truncate_seq; /* last truncate to smaller size */ |
374 | u64 i_truncate_size; /* and the size we last truncated down to */ | |
375 | int i_truncate_pending; /* still need to call vmtruncate */ | |
376 | ||
377 | u64 i_max_size; /* max file size authorized by mds */ | |
378 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ | |
379 | u64 i_wanted_max_size; /* offset we'd like to write too */ | |
380 | u64 i_requested_max_size; /* max_size we've requested */ | |
381 | ||
382 | /* held references to caps */ | |
383 | int i_pin_ref; | |
f85122af | 384 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref; |
de57606c | 385 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
89aa5930 | 386 | atomic_t i_filelock_ref; |
97aeb6bf | 387 | atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ |
cd045cb4 | 388 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
de57606c SW |
389 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
390 | ||
de57606c | 391 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
68cd5b4b | 392 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
de57606c SW |
393 | spinlock_t i_unsafe_lock; |
394 | ||
75c9627e YZ |
395 | union { |
396 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ | |
397 | struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ | |
398 | }; | |
de57606c SW |
399 | int i_snap_realm_counter; /* snap realm (if caps) */ |
400 | struct list_head i_snap_realm_item; | |
401 | struct list_head i_snap_flush_item; | |
245ce991 | 402 | struct timespec64 i_btime; |
193e7b37 | 403 | struct timespec64 i_snap_btime; |
de57606c | 404 | |
1cf89a8d YZ |
405 | struct work_struct i_work; |
406 | unsigned long i_work_mask; | |
de57606c | 407 | |
99ccbd22 MT |
408 | #ifdef CONFIG_CEPH_FSCACHE |
409 | struct fscache_cookie *fscache; | |
f7f7e7a0 | 410 | u32 i_fscache_gen; |
99ccbd22 | 411 | #endif |
f4b97866 YZ |
412 | errseq_t i_meta_err; |
413 | ||
de57606c SW |
414 | struct inode vfs_inode; /* at end */ |
415 | }; | |
416 | ||
721d5c13 JL |
417 | static inline struct ceph_inode_info * |
418 | ceph_inode(const struct inode *inode) | |
de57606c | 419 | { |
fbbccec9 | 420 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
de57606c SW |
421 | } |
422 | ||
721d5c13 JL |
423 | static inline struct ceph_fs_client * |
424 | ceph_inode_to_client(const struct inode *inode) | |
ad1fee96 YS |
425 | { |
426 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; | |
427 | } | |
428 | ||
721d5c13 JL |
429 | static inline struct ceph_fs_client * |
430 | ceph_sb_to_client(const struct super_block *sb) | |
ad1fee96 YS |
431 | { |
432 | return (struct ceph_fs_client *)sb->s_fs_info; | |
433 | } | |
434 | ||
721d5c13 JL |
435 | static inline struct ceph_vino |
436 | ceph_vino(const struct inode *inode) | |
3d14c5d2 YS |
437 | { |
438 | return ceph_inode(inode)->i_vino; | |
439 | } | |
440 | ||
441 | /* | |
442 | * ino_t is <64 bits on many architectures, blech. | |
443 | * | |
ad1fee96 YS |
444 | * i_ino (kernel inode) st_ino (userspace) |
445 | * i386 32 32 | |
446 | * x86_64+ino32 64 32 | |
447 | * x86_64 64 64 | |
448 | */ | |
3310f754 | 449 | static inline u32 ceph_ino_to_ino32(__u64 vino) |
ad1fee96 | 450 | { |
3310f754 AO |
451 | u32 ino = vino & 0xffffffff; |
452 | ino ^= vino >> 32; | |
ad1fee96 | 453 | if (!ino) |
a661fc56 | 454 | ino = 2; |
ad1fee96 YS |
455 | return ino; |
456 | } | |
457 | ||
458 | /* | |
459 | * kernel i_ino value | |
3d14c5d2 YS |
460 | */ |
461 | static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) | |
462 | { | |
3d14c5d2 | 463 | #if BITS_PER_LONG == 32 |
3310f754 AO |
464 | return ceph_ino_to_ino32(vino.ino); |
465 | #else | |
466 | return (ino_t)vino.ino; | |
3d14c5d2 | 467 | #endif |
3d14c5d2 YS |
468 | } |
469 | ||
ad1fee96 YS |
470 | /* |
471 | * user-visible ino (stat, filldir) | |
472 | */ | |
473 | #if BITS_PER_LONG == 32 | |
474 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
475 | { | |
476 | return ino; | |
477 | } | |
478 | #else | |
479 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) | |
480 | { | |
481 | if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) | |
482 | ino = ceph_ino_to_ino32(ino); | |
483 | return ino; | |
484 | } | |
485 | #endif | |
486 | ||
487 | ||
3d14c5d2 YS |
488 | /* for printf-style formatting */ |
489 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap | |
490 | ||
491 | static inline u64 ceph_ino(struct inode *inode) | |
492 | { | |
493 | return ceph_inode(inode)->i_vino.ino; | |
494 | } | |
495 | static inline u64 ceph_snap(struct inode *inode) | |
496 | { | |
497 | return ceph_inode(inode)->i_vino.snap; | |
498 | } | |
499 | ||
500 | static inline int ceph_ino_compare(struct inode *inode, void *data) | |
501 | { | |
502 | struct ceph_vino *pvino = (struct ceph_vino *)data; | |
503 | struct ceph_inode_info *ci = ceph_inode(inode); | |
504 | return ci->i_vino.ino == pvino->ino && | |
505 | ci->i_vino.snap == pvino->snap; | |
506 | } | |
507 | ||
508 | static inline struct inode *ceph_find_inode(struct super_block *sb, | |
509 | struct ceph_vino vino) | |
510 | { | |
511 | ino_t t = ceph_vino_to_ino(vino); | |
512 | return ilookup5(sb, t, ceph_ino_compare, &vino); | |
513 | } | |
514 | ||
515 | ||
516 | /* | |
517 | * Ceph inode. | |
518 | */ | |
10183a69 | 519 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
10183a69 | 520 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ |
daca8bda JL |
521 | #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */ |
522 | #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */ | |
523 | #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */ | |
524 | #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */ | |
c0e385b1 YZ |
525 | #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */ |
526 | #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */ | |
527 | #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */ | |
528 | #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */ | |
529 | #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */ | |
530 | #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */ | |
891f3f5a | 531 | #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT) |
26544c62 | 532 | |
1cf89a8d YZ |
533 | /* |
534 | * Masks of ceph inode work. | |
535 | */ | |
536 | #define CEPH_I_WORK_WRITEBACK 0 /* writeback */ | |
537 | #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */ | |
538 | #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */ | |
539 | ||
26544c62 JL |
540 | /* |
541 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode | |
542 | * and then clear it when they start succeeding. Note that we do a lockless | |
543 | * check first, and only take the lock if it looks like it needs to be changed. | |
544 | * The write submission code just takes this as a hint, so we're not too | |
545 | * worried if a few slip through in either direction. | |
546 | */ | |
547 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) | |
548 | { | |
549 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { | |
550 | spin_lock(&ci->i_ceph_lock); | |
551 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; | |
552 | spin_unlock(&ci->i_ceph_lock); | |
553 | } | |
554 | } | |
555 | ||
556 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) | |
557 | { | |
558 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { | |
559 | spin_lock(&ci->i_ceph_lock); | |
560 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; | |
561 | spin_unlock(&ci->i_ceph_lock); | |
562 | } | |
563 | } | |
3d14c5d2 | 564 | |
2f276c51 | 565 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
fdd4e158 YZ |
566 | long long release_count, |
567 | long long ordered_count) | |
de57606c | 568 | { |
74960773 AP |
569 | /* |
570 | * Makes sure operations that setup readdir cache (update page | |
571 | * cache and i_size) are strongly ordered w.r.t. the following | |
572 | * atomic64_set() operations. | |
573 | */ | |
574 | smp_mb(); | |
fdd4e158 YZ |
575 | atomic64_set(&ci->i_complete_seq[0], release_count); |
576 | atomic64_set(&ci->i_complete_seq[1], ordered_count); | |
de57606c SW |
577 | } |
578 | ||
2f276c51 | 579 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
de57606c | 580 | { |
fdd4e158 YZ |
581 | atomic64_inc(&ci->i_release_count); |
582 | } | |
583 | ||
584 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) | |
585 | { | |
586 | atomic64_inc(&ci->i_ordered_count); | |
2f276c51 | 587 | } |
de57606c | 588 | |
2f276c51 YZ |
589 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
590 | { | |
fdd4e158 YZ |
591 | return atomic64_read(&ci->i_complete_seq[0]) == |
592 | atomic64_read(&ci->i_release_count); | |
de57606c SW |
593 | } |
594 | ||
70db4f36 YZ |
595 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
596 | { | |
fdd4e158 YZ |
597 | return atomic64_read(&ci->i_complete_seq[0]) == |
598 | atomic64_read(&ci->i_release_count) && | |
599 | atomic64_read(&ci->i_complete_seq[1]) == | |
600 | atomic64_read(&ci->i_ordered_count); | |
70db4f36 YZ |
601 | } |
602 | ||
2f276c51 | 603 | static inline void ceph_dir_clear_complete(struct inode *inode) |
de57606c | 604 | { |
2f276c51 YZ |
605 | __ceph_dir_clear_complete(ceph_inode(inode)); |
606 | } | |
de57606c | 607 | |
70db4f36 | 608 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
2f276c51 | 609 | { |
fdd4e158 | 610 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
de57606c SW |
611 | } |
612 | ||
70db4f36 YZ |
613 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
614 | { | |
fdd4e158 YZ |
615 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
616 | smp_rmb(); | |
70db4f36 YZ |
617 | return ret; |
618 | } | |
de57606c SW |
619 | |
620 | /* find a specific frag @f */ | |
621 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, | |
622 | u32 f); | |
623 | ||
624 | /* | |
625 | * choose fragment for value @v. copy frag content to pfrag, if leaf | |
626 | * exists | |
627 | */ | |
628 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, | |
629 | struct ceph_inode_frag *pfrag, | |
630 | int *found); | |
631 | ||
1e9c2eb6 | 632 | static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) |
de57606c SW |
633 | { |
634 | return (struct ceph_dentry_info *)dentry->d_fsdata; | |
635 | } | |
636 | ||
de57606c SW |
637 | /* |
638 | * caps helpers | |
639 | */ | |
640 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) | |
641 | { | |
642 | return !RB_EMPTY_ROOT(&ci->i_caps); | |
643 | } | |
644 | ||
645 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); | |
646 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); | |
647 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, | |
648 | struct ceph_cap *cap); | |
649 | ||
650 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) | |
651 | { | |
652 | int issued; | |
be655596 | 653 | spin_lock(&ci->i_ceph_lock); |
de57606c | 654 | issued = __ceph_caps_issued(ci, NULL); |
be655596 | 655 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
656 | return issued; |
657 | } | |
658 | ||
659 | static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, | |
660 | int touch) | |
661 | { | |
662 | int r; | |
be655596 | 663 | spin_lock(&ci->i_ceph_lock); |
de57606c | 664 | r = __ceph_caps_issued_mask(ci, mask, touch); |
be655596 | 665 | spin_unlock(&ci->i_ceph_lock); |
de57606c SW |
666 | return r; |
667 | } | |
668 | ||
669 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) | |
670 | { | |
671 | return ci->i_dirty_caps | ci->i_flushing_caps; | |
672 | } | |
f66fd9f0 YZ |
673 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
674 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); | |
675 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, | |
676 | struct ceph_cap_flush **pcf); | |
de57606c | 677 | |
9563f88c YZ |
678 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
679 | struct ceph_cap *ocap, int mask); | |
de57606c SW |
680 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
681 | extern int __ceph_caps_used(struct ceph_inode_info *ci); | |
682 | ||
719a2514 YZ |
683 | static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci) |
684 | { | |
685 | return ci->i_nr_by_mode[0]; | |
686 | } | |
de57606c | 687 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); |
525d15e8 | 688 | extern int __ceph_caps_wanted(struct ceph_inode_info *ci); |
de57606c SW |
689 | |
690 | /* what the mds thinks we want */ | |
c1944fed | 691 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
de57606c | 692 | |
37151668 YS |
693 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
694 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); | |
fe33032d YZ |
695 | extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, |
696 | struct ceph_mount_options *fsopt); | |
e30ee581 | 697 | extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 698 | struct ceph_cap_reservation *ctx, int need); |
7bf8f736 | 699 | extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, |
37151668 | 700 | struct ceph_cap_reservation *ctx); |
3d14c5d2 | 701 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
de57606c | 702 | int *total, int *avail, int *used, |
85ccce43 | 703 | int *reserved, int *min); |
de57606c | 704 | |
de57606c | 705 | |
de57606c SW |
706 | |
707 | /* | |
708 | * we keep buffered readdir results attached to file->private_data | |
709 | */ | |
4918b6d1 | 710 | #define CEPH_F_SYNC 1 |
9cfa1098 | 711 | #define CEPH_F_ATEND 2 |
4918b6d1 | 712 | |
de57606c | 713 | struct ceph_file_info { |
252c6728 SW |
714 | short fmode; /* initialized on open */ |
715 | short flags; /* CEPH_F_* */ | |
de57606c | 716 | |
5d988308 YZ |
717 | spinlock_t rw_contexts_lock; |
718 | struct list_head rw_contexts; | |
f4b97866 YZ |
719 | |
720 | errseq_t meta_err; | |
81f148a9 | 721 | u32 filp_gen; |
ff5d913d | 722 | atomic_t num_locks; |
bb48bd4d CX |
723 | }; |
724 | ||
725 | struct ceph_dir_file_info { | |
726 | struct ceph_file_info file_info; | |
5d988308 | 727 | |
de57606c SW |
728 | /* readdir: position within the dir */ |
729 | u32 frag; | |
730 | struct ceph_mds_request *last_readdir; | |
de57606c SW |
731 | |
732 | /* readdir: position within a frag */ | |
f0494206 | 733 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
de57606c | 734 | char *last_name; /* last entry in previous chunk */ |
fdd4e158 YZ |
735 | long long dir_release_count; |
736 | long long dir_ordered_count; | |
737 | int readdir_cache_idx; | |
de57606c SW |
738 | |
739 | /* used for -o dirstat read() on directory thing */ | |
740 | char *dir_info; | |
741 | int dir_info_len; | |
742 | }; | |
743 | ||
5d988308 YZ |
744 | struct ceph_rw_context { |
745 | struct list_head list; | |
746 | struct task_struct *thread; | |
747 | int caps; | |
748 | }; | |
749 | ||
750 | #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ | |
751 | struct ceph_rw_context _name = { \ | |
752 | .thread = current, \ | |
753 | .caps = _caps, \ | |
754 | } | |
755 | ||
756 | static inline void ceph_add_rw_context(struct ceph_file_info *cf, | |
757 | struct ceph_rw_context *ctx) | |
758 | { | |
759 | spin_lock(&cf->rw_contexts_lock); | |
760 | list_add(&ctx->list, &cf->rw_contexts); | |
761 | spin_unlock(&cf->rw_contexts_lock); | |
762 | } | |
763 | ||
764 | static inline void ceph_del_rw_context(struct ceph_file_info *cf, | |
765 | struct ceph_rw_context *ctx) | |
766 | { | |
767 | spin_lock(&cf->rw_contexts_lock); | |
768 | list_del(&ctx->list); | |
769 | spin_unlock(&cf->rw_contexts_lock); | |
770 | } | |
771 | ||
772 | static inline struct ceph_rw_context* | |
773 | ceph_find_rw_context(struct ceph_file_info *cf) | |
774 | { | |
775 | struct ceph_rw_context *ctx, *found = NULL; | |
776 | spin_lock(&cf->rw_contexts_lock); | |
777 | list_for_each_entry(ctx, &cf->rw_contexts, list) { | |
778 | if (ctx->thread == current) { | |
779 | found = ctx; | |
780 | break; | |
781 | } | |
782 | } | |
783 | spin_unlock(&cf->rw_contexts_lock); | |
784 | return found; | |
785 | } | |
786 | ||
fdd4e158 YZ |
787 | struct ceph_readdir_cache_control { |
788 | struct page *page; | |
789 | struct dentry **dentries; | |
790 | int index; | |
791 | }; | |
de57606c | 792 | |
de57606c SW |
793 | /* |
794 | * A "snap realm" describes a subset of the file hierarchy sharing | |
795 | * the same set of snapshots that apply to it. The realms themselves | |
796 | * are organized into a hierarchy, such that children inherit (some of) | |
797 | * the snapshots of their parents. | |
798 | * | |
799 | * All inodes within the realm that have capabilities are linked into a | |
800 | * per-realm list. | |
801 | */ | |
802 | struct ceph_snap_realm { | |
803 | u64 ino; | |
e3161f17 | 804 | struct inode *inode; |
de57606c | 805 | atomic_t nref; |
a105f00c SW |
806 | struct rb_node node; |
807 | ||
de57606c SW |
808 | u64 created, seq; |
809 | u64 parent_ino; | |
810 | u64 parent_since; /* snapid when our current parent became so */ | |
811 | ||
812 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ | |
aa711ee3 | 813 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
de57606c | 814 | u64 *snaps; /* snaps specific to this realm */ |
aa711ee3 | 815 | u32 num_snaps; |
de57606c SW |
816 | |
817 | struct ceph_snap_realm *parent; | |
818 | struct list_head children; /* list of child realms */ | |
819 | struct list_head child_item; | |
820 | ||
821 | struct list_head empty_item; /* if i have ref==0 */ | |
822 | ||
ae00d4f3 SW |
823 | struct list_head dirty_item; /* if realm needs new context */ |
824 | ||
de57606c SW |
825 | /* the current set of snaps for this realm */ |
826 | struct ceph_snap_context *cached_context; | |
827 | ||
828 | struct list_head inodes_with_caps; | |
829 | spinlock_t inodes_with_caps_lock; | |
830 | }; | |
831 | ||
3d14c5d2 | 832 | static inline int default_congestion_kb(void) |
de57606c | 833 | { |
3d14c5d2 YS |
834 | int congestion_kb; |
835 | ||
836 | /* | |
837 | * Copied from NFS | |
838 | * | |
839 | * congestion size, scale with available memory. | |
840 | * | |
841 | * 64MB: 8192k | |
842 | * 128MB: 11585k | |
843 | * 256MB: 16384k | |
844 | * 512MB: 23170k | |
845 | * 1GB: 32768k | |
846 | * 2GB: 46340k | |
847 | * 4GB: 65536k | |
848 | * 8GB: 92681k | |
849 | * 16GB: 131072k | |
850 | * | |
851 | * This allows larger machines to have larger/more transfers. | |
852 | * Limit the default to 256M | |
853 | */ | |
ca79b0c2 | 854 | congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); |
3d14c5d2 YS |
855 | if (congestion_kb > 256*1024) |
856 | congestion_kb = 256*1024; | |
857 | ||
858 | return congestion_kb; | |
de57606c SW |
859 | } |
860 | ||
861 | ||
d468e729 YZ |
862 | /* super.c */ |
863 | extern int ceph_force_reconnect(struct super_block *sb); | |
de57606c SW |
864 | /* snap.c */ |
865 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, | |
866 | u64 ino); | |
867 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, | |
868 | struct ceph_snap_realm *realm); | |
869 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, | |
870 | struct ceph_snap_realm *realm); | |
871 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, | |
982d6011 YZ |
872 | void *p, void *e, bool deletion, |
873 | struct ceph_snap_realm **realm_ret); | |
de57606c | 874 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
2600d2dd | 875 | struct ceph_mds_session *session, |
de57606c | 876 | struct ceph_msg *msg); |
fc837c8f | 877 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
de57606c SW |
878 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
879 | struct ceph_cap_snap *capsnap); | |
880 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); | |
881 | ||
75c9627e YZ |
882 | extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, |
883 | u64 snap); | |
884 | extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, | |
885 | struct ceph_snapid_map *sm); | |
886 | extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); | |
887 | extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); | |
888 | ||
889 | ||
de57606c SW |
890 | /* |
891 | * a cap_snap is "pending" if it is still awaiting an in-progress | |
892 | * sync write (that may/may not still update size, mtime, etc.). | |
893 | */ | |
894 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) | |
895 | { | |
896 | return !list_empty(&ci->i_cap_snaps) && | |
86056090 YZ |
897 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
898 | ci_item)->writing; | |
de57606c SW |
899 | } |
900 | ||
de57606c | 901 | /* inode.c */ |
966c7160 JL |
902 | struct ceph_mds_reply_info_in; |
903 | struct ceph_mds_reply_dirfrag; | |
904 | ||
de57606c SW |
905 | extern const struct inode_operations ceph_file_iops; |
906 | ||
907 | extern struct inode *ceph_alloc_inode(struct super_block *sb); | |
87bc5b89 | 908 | extern void ceph_evict_inode(struct inode *inode); |
cfa6d412 | 909 | extern void ceph_free_inode(struct inode *inode); |
de57606c SW |
910 | |
911 | extern struct inode *ceph_get_inode(struct super_block *sb, | |
912 | struct ceph_vino vino); | |
913 | extern struct inode *ceph_get_snapdir(struct inode *parent); | |
914 | extern int ceph_fill_file_size(struct inode *inode, int issued, | |
915 | u32 truncate_seq, u64 truncate_size, u64 size); | |
916 | extern void ceph_fill_file_time(struct inode *inode, int issued, | |
9bbeab41 AB |
917 | u64 time_warp_seq, struct timespec64 *ctime, |
918 | struct timespec64 *mtime, | |
919 | struct timespec64 *atime); | |
966c7160 JL |
920 | extern int ceph_fill_inode(struct inode *inode, struct page *locked_page, |
921 | struct ceph_mds_reply_info_in *iinfo, | |
922 | struct ceph_mds_reply_dirfrag *dirinfo, | |
923 | struct ceph_mds_session *session, int cap_fmode, | |
924 | struct ceph_cap_reservation *caps_reservation); | |
de57606c | 925 | extern int ceph_fill_trace(struct super_block *sb, |
f5a03b08 | 926 | struct ceph_mds_request *req); |
de57606c SW |
927 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
928 | struct ceph_mds_session *session); | |
929 | ||
930 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); | |
931 | ||
efb0ca76 | 932 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
b415bf4f | 933 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
3c6f6b79 | 934 | extern void ceph_queue_vmtruncate(struct inode *inode); |
3c6f6b79 SW |
935 | extern void ceph_queue_invalidate(struct inode *inode); |
936 | extern void ceph_queue_writeback(struct inode *inode); | |
3e1d0452 | 937 | extern void ceph_async_iput(struct inode *inode); |
de57606c | 938 | |
01deead0 YZ |
939 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
940 | int mask, bool force); | |
941 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) | |
942 | { | |
943 | return __ceph_do_getattr(inode, NULL, mask, force); | |
944 | } | |
10556cb2 | 945 | extern int ceph_permission(struct inode *inode, int mask); |
a26fecca | 946 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
de57606c | 947 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
a528d35e DH |
948 | extern int ceph_getattr(const struct path *path, struct kstat *stat, |
949 | u32 request_mask, unsigned int flags); | |
de57606c SW |
950 | |
951 | /* xattr.c */ | |
a26fecca | 952 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
7221fe4c | 953 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
de57606c | 954 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
12fe3dda | 955 | extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
de57606c | 956 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); |
b1ee94aa | 957 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
de57606c | 958 | |
5c31e92d YZ |
959 | struct ceph_acl_sec_ctx { |
960 | #ifdef CONFIG_CEPH_FS_POSIX_ACL | |
961 | void *default_acl; | |
962 | void *acl; | |
ac6713cc YZ |
963 | #endif |
964 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL | |
965 | void *sec_ctx; | |
966 | u32 sec_ctxlen; | |
5c31e92d YZ |
967 | #endif |
968 | struct ceph_pagelist *pagelist; | |
969 | }; | |
970 | ||
315f2408 YZ |
971 | #ifdef CONFIG_SECURITY |
972 | extern bool ceph_security_xattr_deadlock(struct inode *in); | |
973 | extern bool ceph_security_xattr_wanted(struct inode *in); | |
974 | #else | |
975 | static inline bool ceph_security_xattr_deadlock(struct inode *in) | |
976 | { | |
977 | return false; | |
978 | } | |
979 | static inline bool ceph_security_xattr_wanted(struct inode *in) | |
980 | { | |
981 | return false; | |
982 | } | |
983 | #endif | |
984 | ||
ac6713cc YZ |
985 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL |
986 | extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
987 | struct ceph_acl_sec_ctx *ctx); | |
668959a5 JL |
988 | static inline void ceph_security_invalidate_secctx(struct inode *inode) |
989 | { | |
990 | security_inode_invalidate_secctx(inode); | |
991 | } | |
ac6713cc YZ |
992 | #else |
993 | static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, | |
994 | struct ceph_acl_sec_ctx *ctx) | |
995 | { | |
996 | return 0; | |
997 | } | |
998 | static inline void ceph_security_invalidate_secctx(struct inode *inode) | |
999 | { | |
1000 | } | |
1001 | #endif | |
1002 | ||
5c31e92d | 1003 | void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); |
7221fe4c | 1004 | |
5c31e92d | 1005 | /* acl.c */ |
7221fe4c GZ |
1006 | #ifdef CONFIG_CEPH_FS_POSIX_ACL |
1007 | ||
1008 | struct posix_acl *ceph_get_acl(struct inode *, int); | |
72466d0b | 1009 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
b1ee94aa | 1010 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d YZ |
1011 | struct ceph_acl_sec_ctx *as_ctx); |
1012 | void ceph_init_inode_acls(struct inode *inode, | |
1013 | struct ceph_acl_sec_ctx *as_ctx); | |
c969d9bf GZ |
1014 | |
1015 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
1016 | { | |
1017 | forget_all_cached_acls(inode); | |
1018 | } | |
7221fe4c GZ |
1019 | |
1020 | #else | |
1021 | ||
1022 | #define ceph_get_acl NULL | |
72466d0b | 1023 | #define ceph_set_acl NULL |
7221fe4c | 1024 | |
b1ee94aa | 1025 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
5c31e92d | 1026 | struct ceph_acl_sec_ctx *as_ctx) |
7221fe4c GZ |
1027 | { |
1028 | return 0; | |
1029 | } | |
b1ee94aa | 1030 | static inline void ceph_init_inode_acls(struct inode *inode, |
5c31e92d | 1031 | struct ceph_acl_sec_ctx *as_ctx) |
b1ee94aa YZ |
1032 | { |
1033 | } | |
7221fe4c GZ |
1034 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
1035 | { | |
1036 | return 0; | |
1037 | } | |
1038 | ||
1039 | static inline void ceph_forget_all_cached_acls(struct inode *inode) | |
1040 | { | |
1041 | } | |
1042 | ||
1043 | #endif | |
1044 | ||
de57606c SW |
1045 | /* caps.c */ |
1046 | extern const char *ceph_cap_string(int c); | |
1047 | extern void ceph_handle_caps(struct ceph_mds_session *session, | |
1048 | struct ceph_msg *msg); | |
d9df2783 YZ |
1049 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
1050 | struct ceph_cap_reservation *ctx); | |
1051 | extern void ceph_add_cap(struct inode *inode, | |
1052 | struct ceph_mds_session *session, u64 cap_id, | |
135e671e | 1053 | unsigned issued, unsigned wanted, |
d9df2783 YZ |
1054 | unsigned cap, unsigned seq, u64 realmino, int flags, |
1055 | struct ceph_cap **new_cap); | |
a096b09a | 1056 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
d6e47819 | 1057 | extern void __ceph_remove_caps(struct ceph_inode_info *ci); |
37151668 YS |
1058 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
1059 | struct ceph_cap *cap); | |
9215aeea | 1060 | extern int ceph_is_any_caps(struct inode *inode); |
de57606c | 1061 | |
f1a3d572 | 1062 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
02c24a82 JB |
1063 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
1064 | int datasync); | |
e548e9b9 YZ |
1065 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1066 | struct ceph_mds_session *session); | |
de57606c SW |
1067 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
1068 | struct ceph_mds_session *session); | |
e8a4d267 JL |
1069 | void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, |
1070 | struct ceph_inode_info *ci); | |
2bc50259 GF |
1071 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
1072 | int mds); | |
40dcf75e JL |
1073 | extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, |
1074 | bool snap_rwsem_locked); | |
de57606c SW |
1075 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); |
1076 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); | |
1077 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, | |
1078 | struct ceph_snap_context *snapc); | |
ed9b430c YZ |
1079 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
1080 | struct ceph_mds_session **psession); | |
efb0ca76 | 1081 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
de57606c SW |
1082 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
1083 | struct ceph_mds_session *session); | |
afcdaea3 SW |
1084 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
1085 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); | |
6ef0bc6d | 1086 | extern int ceph_drop_caps_for_unlink(struct inode *inode); |
de57606c SW |
1087 | extern int ceph_encode_inode_release(void **p, struct inode *inode, |
1088 | int mds, int drop, int unless, int force); | |
1089 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, | |
ca6c8ae0 | 1090 | struct inode *dir, |
de57606c SW |
1091 | int mds, int drop, int unless); |
1092 | ||
5e3ded1b | 1093 | extern int ceph_get_caps(struct file *filp, int need, int want, |
31c542a1 | 1094 | loff_t endoff, int *got, struct page **pinned_page); |
5e3ded1b | 1095 | extern int ceph_try_get_caps(struct inode *inode, |
2ee9dd95 | 1096 | int need, int want, bool nonblock, int *got); |
de57606c SW |
1097 | |
1098 | /* for counting open files by mode */ | |
719a2514 YZ |
1099 | extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count); |
1100 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count); | |
1101 | extern void __ceph_touch_fmode(struct ceph_inode_info *ci, | |
1102 | struct ceph_mds_client *mdsc, int fmode); | |
de57606c SW |
1103 | |
1104 | /* addr.c */ | |
1105 | extern const struct address_space_operations ceph_aops; | |
1106 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); | |
10183a69 | 1107 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
5e3ded1b | 1108 | extern int ceph_pool_perm_check(struct inode *inode, int need); |
10183a69 | 1109 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); |
de57606c SW |
1110 | |
1111 | /* file.c */ | |
1112 | extern const struct file_operations ceph_file_fops; | |
9e0eb85d | 1113 | |
719a2514 | 1114 | extern int ceph_renew_caps(struct inode *inode, int fmode); |
de57606c | 1115 | extern int ceph_open(struct inode *inode, struct file *file); |
5ef50c3b | 1116 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
44907d79 | 1117 | struct file *file, unsigned flags, umode_t mode); |
de57606c | 1118 | extern int ceph_release(struct inode *inode, struct file *filp); |
31c542a1 YZ |
1119 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
1120 | char *data, size_t len); | |
55f2a045 | 1121 | |
de57606c SW |
1122 | /* dir.c */ |
1123 | extern const struct file_operations ceph_dir_fops; | |
38c48b5f | 1124 | extern const struct file_operations ceph_snapdir_fops; |
de57606c | 1125 | extern const struct inode_operations ceph_dir_iops; |
38c48b5f | 1126 | extern const struct inode_operations ceph_snapdir_iops; |
18fc8abd | 1127 | extern const struct dentry_operations ceph_dentry_ops; |
de57606c | 1128 | |
f3c4ebe6 | 1129 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
de57606c | 1130 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
468640e3 SW |
1131 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
1132 | struct dentry *dentry, int err); | |
de57606c SW |
1133 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
1134 | struct dentry *dentry, int err); | |
1135 | ||
37c4efc1 YZ |
1136 | extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); |
1137 | extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); | |
81a6cf2d | 1138 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
37c4efc1 | 1139 | extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); |
e5f86dc3 | 1140 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
fdd4e158 | 1141 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
de57606c | 1142 | |
de57606c SW |
1143 | /* ioctl.c */ |
1144 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); | |
1145 | ||
1146 | /* export.c */ | |
1147 | extern const struct export_operations ceph_export_ops; | |
3886274a | 1148 | struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); |
de57606c | 1149 | |
40819f6f | 1150 | /* locks.c */ |
eb13e832 | 1151 | extern __init void ceph_flock_init(void); |
40819f6f GF |
1152 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
1153 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); | |
1154 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); | |
39be95e9 JS |
1155 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
1156 | struct ceph_filelock *flocks, | |
1157 | int num_fcntl_locks, | |
1158 | int num_flock_locks); | |
1159 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, | |
1160 | struct ceph_pagelist *pagelist, | |
1161 | int num_fcntl_locks, int num_flock_locks); | |
40819f6f | 1162 | |
3d14c5d2 | 1163 | /* debugfs.c */ |
1a829ff2 | 1164 | extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); |
3d14c5d2 YS |
1165 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); |
1166 | ||
fb18a575 | 1167 | /* quota.c */ |
d557c48d LH |
1168 | static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) |
1169 | { | |
1170 | return ci->i_max_files || ci->i_max_bytes; | |
1171 | } | |
1172 | ||
1173 | extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); | |
1174 | ||
1175 | static inline void __ceph_update_quota(struct ceph_inode_info *ci, | |
1176 | u64 max_bytes, u64 max_files) | |
1177 | { | |
1178 | bool had_quota, has_quota; | |
1179 | had_quota = __ceph_has_any_quota(ci); | |
1180 | ci->i_max_bytes = max_bytes; | |
1181 | ci->i_max_files = max_files; | |
1182 | has_quota = __ceph_has_any_quota(ci); | |
1183 | ||
1184 | if (had_quota != has_quota) | |
1185 | ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); | |
1186 | } | |
1187 | ||
fb18a575 LH |
1188 | extern void ceph_handle_quota(struct ceph_mds_client *mdsc, |
1189 | struct ceph_mds_session *session, | |
1190 | struct ceph_msg *msg); | |
b7a29217 | 1191 | extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); |
cafe21a4 | 1192 | extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); |
2b83845f LH |
1193 | extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, |
1194 | loff_t newlen); | |
1ab302a0 LH |
1195 | extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, |
1196 | loff_t newlen); | |
9122eed5 LH |
1197 | extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, |
1198 | struct kstatfs *buf); | |
0c44a8e0 | 1199 | extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); |
fb18a575 | 1200 | |
de57606c | 1201 | #endif /* _FS_CEPH_SUPER_H */ |