blob: e705c4d612d763515bb3a85f603c1c7e52107cc7 [file] [log] [blame]
Sage Weilde576062009-10-06 11:31:07 -07001#ifndef _FS_CEPH_SUPER_H
2#define _FS_CEPH_SUPER_H
3
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07004#include <linux/ceph/ceph_debug.h>
Sage Weilde576062009-10-06 11:31:07 -07005
6#include <asm/unaligned.h>
7#include <linux/backing-dev.h>
8#include <linux/completion.h>
9#include <linux/exportfs.h>
10#include <linux/fs.h>
11#include <linux/mempool.h>
12#include <linux/pagemap.h>
13#include <linux/wait.h>
Stephen Rothwellf1a3d572010-01-18 11:53:08 +110014#include <linux/writeback.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090015#include <linux/slab.h>
Guangliang Zhaoc969d9b2014-02-16 08:35:52 -080016#include <linux/posix_acl.h>
Sage Weilde576062009-10-06 11:31:07 -070017
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070018#include <linux/ceph/libceph.h>
Sage Weilde576062009-10-06 11:31:07 -070019
Milosz Tanski99ccbd22013-08-21 17:29:54 -040020#ifdef CONFIG_CEPH_FSCACHE
21#include <linux/fscache.h>
22#endif
23
Sage Weilde576062009-10-06 11:31:07 -070024/* f_type in struct statfs */
25#define CEPH_SUPER_MAGIC 0x00c36400
26
27/* large granularity for statfs utilization stats to facilitate
28 * large volume sizes on 32-bit machines. */
Sage Weil92a49fb2013-02-22 15:31:00 -080029#define CEPH_BLOCK_SHIFT 22 /* 4 MB */
Sage Weilde576062009-10-06 11:31:07 -070030#define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT)
31
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070032#define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */
33#define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */
34#define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */
Yehuda Sadehad1fee92011-01-21 16:44:03 -080035#define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */
Sage Weila40dc6c2012-01-10 09:12:55 -080036#define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */
Milosz Tanski99ccbd22013-08-21 17:29:54 -040037#define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */
Yan, Zheng10183a62015-04-27 15:33:28 +080038#define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */
Sage Weil6a259382010-07-07 09:06:08 -070039
Yan, Zheng133e9152016-01-25 10:44:33 +080040#define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE
Sage Weilde576062009-10-06 11:31:07 -070041
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070042#define ceph_set_mount_opt(fsc, opt) \
43 (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt;
44#define ceph_test_mount_opt(fsc, opt) \
45 (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
Sage Weilde576062009-10-06 11:31:07 -070046
Sage Weil83817e32011-08-04 08:03:44 -070047#define CEPH_RSIZE_DEFAULT 0 /* max read size */
48#define CEPH_RASIZE_DEFAULT (8192*1024) /* readahead */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070049#define CEPH_MAX_READDIR_DEFAULT 1024
50#define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024)
51#define CEPH_SNAPDIRNAME_DEFAULT ".snap"
Sage Weilde576062009-10-06 11:31:07 -070052
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070053struct ceph_mount_options {
Sage Weilde576062009-10-06 11:31:07 -070054 int flags;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070055 int sb_flags;
56
Sage Weil83817e32011-08-04 08:03:44 -070057 int wsize; /* max write size */
58 int rsize; /* max read size */
59 int rasize; /* max readahead */
Sage Weil6e19a162010-04-29 16:38:32 -070060 int congestion_kb; /* max writeback in flight */
61 int caps_wanted_delay_min, caps_wanted_delay_max;
62 int cap_release_safety;
Sage Weil23804d92010-05-14 13:06:30 -070063 int max_readdir; /* max readdir result (entires) */
64 int max_readdir_bytes; /* max readdir result (bytes) */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070065
66 /*
67 * everything above this point can be memcmp'd; everything below
68 * is handled in compare_mount_options()
69 */
70
Sage Weilde576062009-10-06 11:31:07 -070071 char *snapdir_name; /* default ".snap" */
Sage Weilde576062009-10-06 11:31:07 -070072};
73
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070074struct ceph_fs_client {
Sage Weilde576062009-10-06 11:31:07 -070075 struct super_block *sb;
76
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070077 struct ceph_mount_options *mount_options;
78 struct ceph_client *client;
79
Sage Weilde576062009-10-06 11:31:07 -070080 unsigned long mount_state;
Sage Weil85ccce42010-02-17 10:02:43 -080081 int min_caps; /* min caps i added */
82
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070083 struct ceph_mds_client *mdsc;
Sage Weilde576062009-10-06 11:31:07 -070084
85 /* writeback */
86 mempool_t *wb_pagevec_pool;
87 struct workqueue_struct *wb_wq;
88 struct workqueue_struct *pg_inv_wq;
89 struct workqueue_struct *trunc_wq;
Yehuda Sadeh2baba252009-12-18 13:51:57 -080090 atomic_long_t writeback_count;
Sage Weilde576062009-10-06 11:31:07 -070091
92 struct backing_dev_info backing_dev_info;
Sage Weil07433042009-11-18 16:50:41 -080093
94#ifdef CONFIG_DEBUG_FS
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070095 struct dentry *debugfs_dentry_lru, *debugfs_caps;
Yehuda Sadeh2baba252009-12-18 13:51:57 -080096 struct dentry *debugfs_congestion_kb;
Sage Weil06edf042009-12-15 14:44:32 -080097 struct dentry *debugfs_bdi;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070098 struct dentry *debugfs_mdsc, *debugfs_mdsmap;
John Spray14ed9702014-09-12 16:58:49 +010099 struct dentry *debugfs_mds_sessions;
Sage Weil07433042009-11-18 16:50:41 -0800100#endif
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400101
102#ifdef CONFIG_CEPH_FSCACHE
103 struct fscache_cookie *fscache;
104 struct workqueue_struct *revalidate_wq;
105#endif
Sage Weilde576062009-10-06 11:31:07 -0700106};
107
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700108
Sage Weilde576062009-10-06 11:31:07 -0700109/*
110 * File i/o capability. This tracks shared state with the metadata
111 * server that allows us to cache or writeback attributes or to read
112 * and write data. For any given inode, we should have one or more
113 * capabilities, one issued by each metadata server, and our
114 * cumulative access is the OR of all issued capabilities.
115 *
116 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
117 * session capability lists.
118 */
119struct ceph_cap {
120 struct ceph_inode_info *ci;
121 struct rb_node ci_node; /* per-ci cap tree */
122 struct ceph_mds_session *session;
123 struct list_head session_caps; /* per-session caplist */
Sage Weilde576062009-10-06 11:31:07 -0700124 u64 cap_id; /* unique cap id (mds provided) */
Yan, Zheng745a8e32015-05-14 17:22:42 +0800125 union {
126 /* in-use caps */
127 struct {
128 int issued; /* latest, from the mds */
129 int implemented; /* implemented superset of
130 issued (for revocation) */
131 int mds, mds_wanted;
132 };
133 /* caps to release */
134 struct {
135 u64 cap_ino;
136 int queue_release;
137 };
138 };
Sage Weil685f9a5d2009-11-09 12:05:48 -0800139 u32 seq, issue_seq, mseq;
140 u32 cap_gen; /* active/stale cycle */
Sage Weilde576062009-10-06 11:31:07 -0700141 unsigned long last_used;
142 struct list_head caps_item;
143};
144
145#define CHECK_CAPS_NODELAY 1 /* do not delay any further */
146#define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */
147#define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */
148
149/*
150 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
151 * we first complete any in-process sync writes and writeback any dirty
152 * data before flushing the snapped state (tracked here) back to the MDS.
153 */
154struct ceph_cap_snap {
155 atomic_t nref;
156 struct ceph_inode_info *ci;
157 struct list_head ci_item, flushing_item;
158
159 u64 follows, flush_tid;
160 int issued, dirty;
161 struct ceph_snap_context *context;
162
Al Viro5706b272011-07-26 04:52:22 -0400163 umode_t mode;
Eric W. Biederman05cb11c2013-01-31 02:56:19 -0800164 kuid_t uid;
165 kgid_t gid;
Sage Weilde576062009-10-06 11:31:07 -0700166
Sage Weil4a625be2010-08-22 15:03:56 -0700167 struct ceph_buffer *xattr_blob;
Sage Weilde576062009-10-06 11:31:07 -0700168 u64 xattr_version;
169
170 u64 size;
171 struct timespec mtime, atime, ctime;
172 u64 time_warp_seq;
173 int writing; /* a sync write is still in progress */
174 int dirty_pages; /* dirty pages awaiting writeback */
Yan, Zhenge20d2582014-11-14 22:39:13 +0800175 bool inline_data;
Yan, Zheng86056092015-05-01 16:57:16 +0800176 bool need_flush;
Sage Weilde576062009-10-06 11:31:07 -0700177};
178
179static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
180{
Sage Weil4a625be2010-08-22 15:03:56 -0700181 if (atomic_dec_and_test(&capsnap->nref)) {
182 if (capsnap->xattr_blob)
183 ceph_buffer_put(capsnap->xattr_blob);
Sage Weilde576062009-10-06 11:31:07 -0700184 kfree(capsnap);
Sage Weil4a625be2010-08-22 15:03:56 -0700185 }
Sage Weilde576062009-10-06 11:31:07 -0700186}
187
Yan, Zheng553adfd2015-06-09 15:48:57 +0800188struct ceph_cap_flush {
189 u64 tid;
190 int caps;
Yan, Zhenge548e9b2015-06-10 15:17:56 +0800191 struct rb_node g_node; // global
Yan, Zheng553adfd2015-06-09 15:48:57 +0800192 union {
Yan, Zhenge548e9b2015-06-10 15:17:56 +0800193 struct rb_node i_node; // inode
Yan, Zheng553adfd2015-06-09 15:48:57 +0800194 struct list_head list;
195 };
196};
197
Sage Weilde576062009-10-06 11:31:07 -0700198/*
199 * The frag tree describes how a directory is fragmented, potentially across
200 * multiple metadata servers. It is also used to indicate points where
201 * metadata authority is delegated, and whether/where metadata is replicated.
202 *
203 * A _leaf_ frag will be present in the i_fragtree IFF there is
204 * delegation info. That is, if mds >= 0 || ndist > 0.
205 */
206#define CEPH_MAX_DIRFRAG_REP 4
207
208struct ceph_inode_frag {
209 struct rb_node node;
210
211 /* fragtree state */
212 u32 frag;
213 int split_by; /* i.e. 2^(split_by) children */
214
215 /* delegation and replication info */
216 int mds; /* -1 if same authority as parent */
217 int ndist; /* >0 if replicated */
218 int dist[CEPH_MAX_DIRFRAG_REP];
219};
220
221/*
222 * We cache inode xattrs as an encoded blob until they are first used,
223 * at which point we parse them into an rbtree.
224 */
225struct ceph_inode_xattr {
226 struct rb_node node;
227
228 const char *name;
229 int name_len;
230 const char *val;
231 int val_len;
232 int dirty;
233
234 int should_free_name;
235 int should_free_val;
236};
237
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700238/*
239 * Ceph dentry state
240 */
241struct ceph_dentry_info {
242 struct ceph_mds_session *lease_session;
243 u32 lease_gen, lease_shared_gen;
244 u32 lease_seq;
245 unsigned long lease_renew_after, lease_renew_from;
246 struct list_head lru;
247 struct dentry *dentry;
248 u64 time;
249 u64 offset;
250};
251
Sage Weilde576062009-10-06 11:31:07 -0700252struct ceph_inode_xattrs_info {
253 /*
254 * (still encoded) xattr blob. we avoid the overhead of parsing
255 * this until someone actually calls getxattr, etc.
256 *
257 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
258 * NULL means we don't know.
259 */
260 struct ceph_buffer *blob, *prealloc_blob;
261
262 struct rb_root index;
263 bool dirty;
264 int count;
265 int names_size;
266 int vals_size;
267 u64 version, index_version;
268};
269
270/*
271 * Ceph inode.
272 */
Sage Weilde576062009-10-06 11:31:07 -0700273struct ceph_inode_info {
274 struct ceph_vino i_vino; /* ceph ino + snap */
275
Sage Weilbe655592011-11-30 09:47:09 -0800276 spinlock_t i_ceph_lock;
277
Sage Weilde576062009-10-06 11:31:07 -0700278 u64 i_version;
Yan, Zheng31c542a2014-11-14 21:41:55 +0800279 u64 i_inline_version;
Sage Weilde576062009-10-06 11:31:07 -0700280 u32 i_time_warp_seq;
281
282 unsigned i_ceph_flags;
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800283 atomic64_t i_release_count;
284 atomic64_t i_ordered_count;
285 atomic64_t i_complete_seq[2];
Sage Weilde576062009-10-06 11:31:07 -0700286
Sage Weil6c0f3af2010-11-16 11:14:34 -0800287 struct ceph_dir_layout i_dir_layout;
Sage Weilde576062009-10-06 11:31:07 -0700288 struct ceph_file_layout i_layout;
Yan, Zheng5ea5c5e2016-02-14 18:06:41 +0800289 size_t i_pool_ns_len;
Sage Weilde576062009-10-06 11:31:07 -0700290 char *i_symlink;
291
292 /* for dirs */
293 struct timespec i_rctime;
294 u64 i_rbytes, i_rfiles, i_rsubdirs;
295 u64 i_files, i_subdirs;
Sage Weilde576062009-10-06 11:31:07 -0700296
297 struct rb_root i_fragtree;
298 struct mutex i_fragtree_mutex;
299
300 struct ceph_inode_xattrs_info i_xattrs;
301
Sage Weilbe655592011-11-30 09:47:09 -0800302 /* capabilities. protected _both_ by i_ceph_lock and cap->session's
Sage Weilde576062009-10-06 11:31:07 -0700303 * s_mutex. */
304 struct rb_root i_caps; /* cap list */
305 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
306 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
307 struct list_head i_dirty_item, i_flushing_item;
Sage Weilde576062009-10-06 11:31:07 -0700308 /* we need to track cap writeback on a per-cap-bit basis, to allow
309 * overlapping, pipelined cap flushes to the mds. we can probably
310 * reduce the tid to 8 bits if we're concerned about inode size. */
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800311 struct ceph_cap_flush *i_prealloc_cap_flush;
Yan, Zheng553adfd2015-06-09 15:48:57 +0800312 struct rb_root i_cap_flush_tree;
Sage Weilde576062009-10-06 11:31:07 -0700313 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
314 unsigned long i_hold_caps_min; /* jiffies */
315 unsigned long i_hold_caps_max; /* jiffies */
316 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
Sage Weilde576062009-10-06 11:31:07 -0700317 struct ceph_cap_reservation i_cap_migration_resv;
318 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
Sage Weil7d8cb262010-08-24 08:44:16 -0700319 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or
320 dirty|flushing caps */
Sage Weilde576062009-10-06 11:31:07 -0700321 unsigned i_snap_caps; /* cap bits for snapped files */
322
323 int i_nr_by_mode[CEPH_FILE_MODE_NUM]; /* open file counts */
324
Yan, Zhengb0d7c222013-08-12 21:42:15 -0700325 struct mutex i_truncate_mutex;
Sage Weilde576062009-10-06 11:31:07 -0700326 u32 i_truncate_seq; /* last truncate to smaller size */
327 u64 i_truncate_size; /* and the size we last truncated down to */
328 int i_truncate_pending; /* still need to call vmtruncate */
329
330 u64 i_max_size; /* max file size authorized by mds */
331 u64 i_reported_size; /* (max_)size reported to or requested of mds */
332 u64 i_wanted_max_size; /* offset we'd like to write too */
333 u64 i_requested_max_size; /* max_size we've requested */
334
335 /* held references to caps */
336 int i_pin_ref;
Henry C Changd3d07202011-05-11 10:29:54 +0000337 int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref;
Sage Weilde576062009-10-06 11:31:07 -0700338 int i_wrbuffer_ref, i_wrbuffer_ref_head;
339 u32 i_shared_gen; /* increment each time we get FILE_SHARED */
Sage Weilcd045cb2010-11-04 11:05:05 -0700340 u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */
Sage Weilde576062009-10-06 11:31:07 -0700341 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
342
343 struct list_head i_unsafe_writes; /* uncommitted sync writes */
344 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
Yan, Zheng68cd5b42015-10-27 18:36:06 +0800345 struct list_head i_unsafe_iops; /* uncommitted mds inode ops */
Sage Weilde576062009-10-06 11:31:07 -0700346 spinlock_t i_unsafe_lock;
347
348 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
349 int i_snap_realm_counter; /* snap realm (if caps) */
350 struct list_head i_snap_realm_item;
351 struct list_head i_snap_flush_item;
352
353 struct work_struct i_wb_work; /* writeback work */
354 struct work_struct i_pg_inv_work; /* page invalidation work */
355
356 struct work_struct i_vmtruncate_work;
357
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400358#ifdef CONFIG_CEPH_FSCACHE
359 struct fscache_cookie *fscache;
360 u32 i_fscache_gen; /* sequence, for delayed fscache validate */
361 struct work_struct i_revalidate_work;
362#endif
Sage Weilde576062009-10-06 11:31:07 -0700363 struct inode vfs_inode; /* at end */
364};
365
366static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
367{
Noah Watkinsfbbccec2009-10-28 11:54:49 -0700368 return container_of(inode, struct ceph_inode_info, vfs_inode);
Sage Weilde576062009-10-06 11:31:07 -0700369}
370
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800371static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode)
372{
373 return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
374}
375
376static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb)
377{
378 return (struct ceph_fs_client *)sb->s_fs_info;
379}
380
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700381static inline struct ceph_vino ceph_vino(struct inode *inode)
382{
383 return ceph_inode(inode)->i_vino;
384}
385
386/*
387 * ino_t is <64 bits on many architectures, blech.
388 *
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800389 * i_ino (kernel inode) st_ino (userspace)
390 * i386 32 32
391 * x86_64+ino32 64 32
392 * x86_64 64 64
393 */
Amon Ott3310f752011-10-20 13:04:07 -0700394static inline u32 ceph_ino_to_ino32(__u64 vino)
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800395{
Amon Ott3310f752011-10-20 13:04:07 -0700396 u32 ino = vino & 0xffffffff;
397 ino ^= vino >> 32;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800398 if (!ino)
Amon Otta661fc52012-01-23 09:25:23 -0800399 ino = 2;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800400 return ino;
401}
402
403/*
404 * kernel i_ino value
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700405 */
406static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
407{
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700408#if BITS_PER_LONG == 32
Amon Ott3310f752011-10-20 13:04:07 -0700409 return ceph_ino_to_ino32(vino.ino);
410#else
411 return (ino_t)vino.ino;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700412#endif
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700413}
414
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800415/*
416 * user-visible ino (stat, filldir)
417 */
418#if BITS_PER_LONG == 32
419static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
420{
421 return ino;
422}
423#else
424static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
425{
426 if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))
427 ino = ceph_ino_to_ino32(ino);
428 return ino;
429}
430#endif
431
432
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700433/* for printf-style formatting */
434#define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
435
436static inline u64 ceph_ino(struct inode *inode)
437{
438 return ceph_inode(inode)->i_vino.ino;
439}
440static inline u64 ceph_snap(struct inode *inode)
441{
442 return ceph_inode(inode)->i_vino.snap;
443}
444
445static inline int ceph_ino_compare(struct inode *inode, void *data)
446{
447 struct ceph_vino *pvino = (struct ceph_vino *)data;
448 struct ceph_inode_info *ci = ceph_inode(inode);
449 return ci->i_vino.ino == pvino->ino &&
450 ci->i_vino.snap == pvino->snap;
451}
452
453static inline struct inode *ceph_find_inode(struct super_block *sb,
454 struct ceph_vino vino)
455{
456 ino_t t = ceph_vino_to_ino(vino);
457 return ilookup5(sb, t, ceph_ino_compare, &vino);
458}
459
460
461/*
462 * Ceph inode.
463 */
Yan, Zheng10183a62015-04-27 15:33:28 +0800464#define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */
465#define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */
466#define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */
467#define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */
468#define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */
469#define CEPH_I_POOL_RD (1 << 5) /* can read from pool */
470#define CEPH_I_POOL_WR (1 << 6) /* can write to pool */
Yan, Zheng315f2402016-03-07 10:34:50 +0800471#define CEPH_I_SEC_INITED (1 << 7) /* security initialized */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700472
Yan, Zheng2f276c52013-03-13 19:44:32 +0800473static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800474 long long release_count,
475 long long ordered_count)
Sage Weilde576062009-10-06 11:31:07 -0700476{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800477 smp_mb__before_atomic();
478 atomic64_set(&ci->i_complete_seq[0], release_count);
479 atomic64_set(&ci->i_complete_seq[1], ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700480}
481
Yan, Zheng2f276c52013-03-13 19:44:32 +0800482static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700483{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800484 atomic64_inc(&ci->i_release_count);
485}
486
487static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
488{
489 atomic64_inc(&ci->i_ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700490}
491
Yan, Zheng2f276c52013-03-13 19:44:32 +0800492static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700493{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800494 return atomic64_read(&ci->i_complete_seq[0]) ==
495 atomic64_read(&ci->i_release_count);
Yan, Zheng2f276c52013-03-13 19:44:32 +0800496}
Sage Weilde576062009-10-06 11:31:07 -0700497
Yan, Zheng70db4f32014-10-21 18:09:56 -0700498static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
499{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800500 return atomic64_read(&ci->i_complete_seq[0]) ==
501 atomic64_read(&ci->i_release_count) &&
502 atomic64_read(&ci->i_complete_seq[1]) ==
503 atomic64_read(&ci->i_ordered_count);
Yan, Zheng70db4f32014-10-21 18:09:56 -0700504}
505
Yan, Zheng2f276c52013-03-13 19:44:32 +0800506static inline void ceph_dir_clear_complete(struct inode *inode)
507{
508 __ceph_dir_clear_complete(ceph_inode(inode));
509}
510
Yan, Zheng70db4f32014-10-21 18:09:56 -0700511static inline void ceph_dir_clear_ordered(struct inode *inode)
Yan, Zheng2f276c52013-03-13 19:44:32 +0800512{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800513 __ceph_dir_clear_ordered(ceph_inode(inode));
Sage Weilde576062009-10-06 11:31:07 -0700514}
515
Yan, Zheng70db4f32014-10-21 18:09:56 -0700516static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
517{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800518 bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
519 smp_rmb();
Yan, Zheng70db4f32014-10-21 18:09:56 -0700520 return ret;
521}
Sage Weilde576062009-10-06 11:31:07 -0700522
523/* find a specific frag @f */
524extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
525 u32 f);
526
527/*
528 * choose fragment for value @v. copy frag content to pfrag, if leaf
529 * exists
530 */
531extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
532 struct ceph_inode_frag *pfrag,
533 int *found);
534
Sage Weilde576062009-10-06 11:31:07 -0700535static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
536{
537 return (struct ceph_dentry_info *)dentry->d_fsdata;
538}
539
540static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
541{
542 return ((loff_t)frag << 32) | (loff_t)off;
543}
544
Sage Weilde576062009-10-06 11:31:07 -0700545/*
546 * caps helpers
547 */
548static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
549{
550 return !RB_EMPTY_ROOT(&ci->i_caps);
551}
552
553extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
554extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
555extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
556 struct ceph_cap *cap);
557
558static inline int ceph_caps_issued(struct ceph_inode_info *ci)
559{
560 int issued;
Sage Weilbe655592011-11-30 09:47:09 -0800561 spin_lock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700562 issued = __ceph_caps_issued(ci, NULL);
Sage Weilbe655592011-11-30 09:47:09 -0800563 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700564 return issued;
565}
566
567static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
568 int touch)
569{
570 int r;
Sage Weilbe655592011-11-30 09:47:09 -0800571 spin_lock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700572 r = __ceph_caps_issued_mask(ci, mask, touch);
Sage Weilbe655592011-11-30 09:47:09 -0800573 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700574 return r;
575}
576
577static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
578{
579 return ci->i_dirty_caps | ci->i_flushing_caps;
580}
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800581extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
582extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
583extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
584 struct ceph_cap_flush **pcf);
Sage Weilde576062009-10-06 11:31:07 -0700585
Yan, Zheng9563f882013-11-22 13:50:45 +0800586extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
587 struct ceph_cap *ocap, int mask);
Sage Weilde576062009-10-06 11:31:07 -0700588extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
589extern int __ceph_caps_used(struct ceph_inode_info *ci);
590
591extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
592
593/*
594 * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
595 */
596static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
597{
598 int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
599 if (w & CEPH_CAP_FILE_BUFFER)
600 w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */
601 return w;
602}
603
604/* what the mds thinks we want */
605extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
606
Yehuda Sadeh37151662010-06-17 16:16:12 -0700607extern void ceph_caps_init(struct ceph_mds_client *mdsc);
608extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
609extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
majianpeng93faca62013-06-26 11:15:27 +0800610extern void ceph_reserve_caps(struct ceph_mds_client *mdsc,
Yehuda Sadeh37151662010-06-17 16:16:12 -0700611 struct ceph_cap_reservation *ctx, int need);
612extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
613 struct ceph_cap_reservation *ctx);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700614extern void ceph_reservation_status(struct ceph_fs_client *client,
Sage Weilde576062009-10-06 11:31:07 -0700615 int *total, int *avail, int *used,
Sage Weil85ccce42010-02-17 10:02:43 -0800616 int *reserved, int *min);
Sage Weilde576062009-10-06 11:31:07 -0700617
Sage Weilde576062009-10-06 11:31:07 -0700618
Sage Weilde576062009-10-06 11:31:07 -0700619
620/*
621 * we keep buffered readdir results attached to file->private_data
622 */
Sage Weil4918b6d2011-07-26 11:26:07 -0700623#define CEPH_F_SYNC 1
Sage Weil9cfa1092011-07-26 11:26:18 -0700624#define CEPH_F_ATEND 2
Sage Weil4918b6d2011-07-26 11:26:07 -0700625
Sage Weilde576062009-10-06 11:31:07 -0700626struct ceph_file_info {
Sage Weil252c6722011-07-26 11:25:27 -0700627 short fmode; /* initialized on open */
628 short flags; /* CEPH_F_* */
Sage Weilde576062009-10-06 11:31:07 -0700629
630 /* readdir: position within the dir */
631 u32 frag;
632 struct ceph_mds_request *last_readdir;
Sage Weilde576062009-10-06 11:31:07 -0700633
634 /* readdir: position within a frag */
635 unsigned offset; /* offset of last chunk, adjusted for . and .. */
Yan, Zhengf0494202014-02-27 16:26:24 +0800636 unsigned next_offset; /* offset of next chunk (last_name's + 1) */
Sage Weilde576062009-10-06 11:31:07 -0700637 char *last_name; /* last entry in previous chunk */
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800638 long long dir_release_count;
639 long long dir_ordered_count;
640 int readdir_cache_idx;
Sage Weilde576062009-10-06 11:31:07 -0700641
642 /* used for -o dirstat read() on directory thing */
643 char *dir_info;
644 int dir_info_len;
645};
646
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800647struct ceph_readdir_cache_control {
648 struct page *page;
649 struct dentry **dentries;
650 int index;
651};
Sage Weilde576062009-10-06 11:31:07 -0700652
653/*
Sage Weilde576062009-10-06 11:31:07 -0700654 * A "snap realm" describes a subset of the file hierarchy sharing
655 * the same set of snapshots that apply to it. The realms themselves
656 * are organized into a hierarchy, such that children inherit (some of)
657 * the snapshots of their parents.
658 *
659 * All inodes within the realm that have capabilities are linked into a
660 * per-realm list.
661 */
662struct ceph_snap_realm {
663 u64 ino;
664 atomic_t nref;
Sage Weila105f002010-02-15 14:37:55 -0800665 struct rb_node node;
666
Sage Weilde576062009-10-06 11:31:07 -0700667 u64 created, seq;
668 u64 parent_ino;
669 u64 parent_since; /* snapid when our current parent became so */
670
671 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
Alex Elderaa711ee2012-07-13 20:35:11 -0500672 u32 num_prior_parent_snaps; /* had prior to parent_since */
Sage Weilde576062009-10-06 11:31:07 -0700673 u64 *snaps; /* snaps specific to this realm */
Alex Elderaa711ee2012-07-13 20:35:11 -0500674 u32 num_snaps;
Sage Weilde576062009-10-06 11:31:07 -0700675
676 struct ceph_snap_realm *parent;
677 struct list_head children; /* list of child realms */
678 struct list_head child_item;
679
680 struct list_head empty_item; /* if i have ref==0 */
681
Sage Weilae00d4f2010-09-16 16:26:51 -0700682 struct list_head dirty_item; /* if realm needs new context */
683
Sage Weilde576062009-10-06 11:31:07 -0700684 /* the current set of snaps for this realm */
685 struct ceph_snap_context *cached_context;
686
687 struct list_head inodes_with_caps;
688 spinlock_t inodes_with_caps_lock;
689};
690
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700691static inline int default_congestion_kb(void)
Sage Weilde576062009-10-06 11:31:07 -0700692{
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700693 int congestion_kb;
694
695 /*
696 * Copied from NFS
697 *
698 * congestion size, scale with available memory.
699 *
700 * 64MB: 8192k
701 * 128MB: 11585k
702 * 256MB: 16384k
703 * 512MB: 23170k
704 * 1GB: 32768k
705 * 2GB: 46340k
706 * 4GB: 65536k
707 * 8GB: 92681k
708 * 16GB: 131072k
709 *
710 * This allows larger machines to have larger/more transfers.
711 * Limit the default to 256M
712 */
713 congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
714 if (congestion_kb > 256*1024)
715 congestion_kb = 256*1024;
716
717 return congestion_kb;
Sage Weilde576062009-10-06 11:31:07 -0700718}
719
720
721
722/* snap.c */
723struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
724 u64 ino);
725extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
726 struct ceph_snap_realm *realm);
727extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
728 struct ceph_snap_realm *realm);
729extern int ceph_update_snap_trace(struct ceph_mds_client *m,
Yan, Zheng982d6012014-12-23 15:30:54 +0800730 void *p, void *e, bool deletion,
731 struct ceph_snap_realm **realm_ret);
Sage Weilde576062009-10-06 11:31:07 -0700732extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
Sage Weil2600d2d2010-02-22 15:12:16 -0800733 struct ceph_mds_session *session,
Sage Weilde576062009-10-06 11:31:07 -0700734 struct ceph_msg *msg);
Sage Weilfc837c8f2010-04-13 11:41:22 -0700735extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
Sage Weilde576062009-10-06 11:31:07 -0700736extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
737 struct ceph_cap_snap *capsnap);
738extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
739
740/*
741 * a cap_snap is "pending" if it is still awaiting an in-progress
742 * sync write (that may/may not still update size, mtime, etc.).
743 */
744static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
745{
746 return !list_empty(&ci->i_cap_snaps) &&
Yan, Zheng86056092015-05-01 16:57:16 +0800747 list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
748 ci_item)->writing;
Sage Weilde576062009-10-06 11:31:07 -0700749}
750
Sage Weilde576062009-10-06 11:31:07 -0700751/* inode.c */
752extern const struct inode_operations ceph_file_iops;
753
754extern struct inode *ceph_alloc_inode(struct super_block *sb);
755extern void ceph_destroy_inode(struct inode *inode);
Yan, Zheng9f12bd12013-09-20 19:55:31 +0800756extern int ceph_drop_inode(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700757
758extern struct inode *ceph_get_inode(struct super_block *sb,
759 struct ceph_vino vino);
760extern struct inode *ceph_get_snapdir(struct inode *parent);
761extern int ceph_fill_file_size(struct inode *inode, int issued,
762 u32 truncate_seq, u64 truncate_size, u64 size);
763extern void ceph_fill_file_time(struct inode *inode, int issued,
764 u64 time_warp_seq, struct timespec *ctime,
765 struct timespec *mtime, struct timespec *atime);
766extern int ceph_fill_trace(struct super_block *sb,
767 struct ceph_mds_request *req,
768 struct ceph_mds_session *session);
769extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
770 struct ceph_mds_session *session);
771
772extern int ceph_inode_holds_cap(struct inode *inode, int mask);
773
774extern int ceph_inode_set_size(struct inode *inode, loff_t size);
Yan, Zhengb415bf42013-07-02 12:40:19 +0800775extern void __ceph_do_pending_vmtruncate(struct inode *inode);
Sage Weil3c6f6b72010-02-09 15:24:44 -0800776extern void ceph_queue_vmtruncate(struct inode *inode);
777
778extern void ceph_queue_invalidate(struct inode *inode);
779extern void ceph_queue_writeback(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700780
Yan, Zheng01deead2014-11-14 21:56:29 +0800781extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
782 int mask, bool force);
783static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
784{
785 return __ceph_do_getattr(inode, NULL, mask, force);
786}
Al Viro10556cb2011-06-20 19:28:19 -0400787extern int ceph_permission(struct inode *inode, int mask);
Sage Weilde576062009-10-06 11:31:07 -0700788extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
789extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
790 struct kstat *stat);
791
792/* xattr.c */
793extern int ceph_setxattr(struct dentry *, const char *, const void *,
794 size_t, int);
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800795int __ceph_setxattr(struct dentry *, const char *, const void *, size_t, int);
796ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
797int __ceph_removexattr(struct dentry *, const char *);
Sage Weilde576062009-10-06 11:31:07 -0700798extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t);
799extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
800extern int ceph_removexattr(struct dentry *, const char *);
801extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
802extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
Alex Elder3ce6cd12012-01-23 15:49:28 -0600803extern void __init ceph_xattr_init(void);
804extern void ceph_xattr_exit(void);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800805extern const struct xattr_handler *ceph_xattr_handlers[];
Sage Weilde576062009-10-06 11:31:07 -0700806
Yan, Zheng315f2402016-03-07 10:34:50 +0800807#ifdef CONFIG_SECURITY
808extern bool ceph_security_xattr_deadlock(struct inode *in);
809extern bool ceph_security_xattr_wanted(struct inode *in);
810#else
811static inline bool ceph_security_xattr_deadlock(struct inode *in)
812{
813 return false;
814}
815static inline bool ceph_security_xattr_wanted(struct inode *in)
816{
817 return false;
818}
819#endif
820
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800821/* acl.c */
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800822struct ceph_acls_info {
823 void *default_acl;
824 void *acl;
825 struct ceph_pagelist *pagelist;
826};
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800827
828#ifdef CONFIG_CEPH_FS_POSIX_ACL
829
830struct posix_acl *ceph_get_acl(struct inode *, int);
Sage Weil72466d02014-01-29 06:22:25 -0800831int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800832int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
833 struct ceph_acls_info *info);
834void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info);
835void ceph_release_acls_info(struct ceph_acls_info *info);
Guangliang Zhaoc969d9b2014-02-16 08:35:52 -0800836
837static inline void ceph_forget_all_cached_acls(struct inode *inode)
838{
839 forget_all_cached_acls(inode);
840}
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800841
842#else
843
844#define ceph_get_acl NULL
Sage Weil72466d02014-01-29 06:22:25 -0800845#define ceph_set_acl NULL
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800846
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800847static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
848 struct ceph_acls_info *info)
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800849{
850 return 0;
851}
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800852static inline void ceph_init_inode_acls(struct inode *inode,
853 struct ceph_acls_info *info)
854{
855}
856static inline void ceph_release_acls_info(struct ceph_acls_info *info)
857{
858}
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800859static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
860{
861 return 0;
862}
863
864static inline void ceph_forget_all_cached_acls(struct inode *inode)
865{
866}
867
868#endif
869
Sage Weilde576062009-10-06 11:31:07 -0700870/* caps.c */
871extern const char *ceph_cap_string(int c);
872extern void ceph_handle_caps(struct ceph_mds_session *session,
873 struct ceph_msg *msg);
Yan, Zhengd9df2782014-04-18 09:57:11 +0800874extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
875 struct ceph_cap_reservation *ctx);
876extern void ceph_add_cap(struct inode *inode,
877 struct ceph_mds_session *session, u64 cap_id,
878 int fmode, unsigned issued, unsigned wanted,
879 unsigned cap, unsigned seq, u64 realmino, int flags,
880 struct ceph_cap **new_cap);
Yan, Zhenga096b092013-09-22 10:15:58 +0800881extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
Yehuda Sadeh37151662010-06-17 16:16:12 -0700882extern void ceph_put_cap(struct ceph_mds_client *mdsc,
883 struct ceph_cap *cap);
Yan, Zheng9215aee2013-11-30 12:47:41 +0800884extern int ceph_is_any_caps(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700885
886extern void ceph_queue_caps_release(struct inode *inode);
Stephen Rothwellf1a3d572010-01-18 11:53:08 +1100887extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
Josef Bacik02c24a82011-07-16 20:44:56 -0400888extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
889 int datasync);
Yan, Zhenge548e9b2015-06-10 15:17:56 +0800890extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
891 struct ceph_mds_session *session);
Sage Weilde576062009-10-06 11:31:07 -0700892extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
893 struct ceph_mds_session *session);
Greg Farnum2bc50252010-06-30 12:44:34 -0700894extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
895 int mds);
Sage Weilde576062009-10-06 11:31:07 -0700896extern int ceph_get_cap_mds(struct inode *inode);
897extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
898extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
899extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
900 struct ceph_snap_context *snapc);
901extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
Sage Weile8351242010-09-17 08:03:08 -0700902 struct ceph_mds_session **psession,
903 int again);
Sage Weilde576062009-10-06 11:31:07 -0700904extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
905 struct ceph_mds_session *session);
Sage Weilafcdaea2009-10-14 14:27:38 -0700906extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
907extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
Sage Weilde576062009-10-06 11:31:07 -0700908
909extern int ceph_encode_inode_release(void **p, struct inode *inode,
910 int mds, int drop, int unless, int force);
911extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
912 int mds, int drop, int unless);
913
914extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
Yan, Zheng31c542a2014-11-14 21:41:55 +0800915 loff_t endoff, int *got, struct page **pinned_page);
Sage Weilde576062009-10-06 11:31:07 -0700916
917/* for counting open files by mode */
918static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
919{
920 ci->i_nr_by_mode[mode]++;
921}
922extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
923
924/* addr.c */
925extern const struct address_space_operations ceph_aops;
926extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
Yan, Zheng10183a62015-04-27 15:33:28 +0800927extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
928extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need);
929extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
Sage Weilde576062009-10-06 11:31:07 -0700930
931/* file.c */
932extern const struct file_operations ceph_file_fops;
Alex Elder9e0eb852013-02-06 13:11:38 -0600933
Sage Weilde576062009-10-06 11:31:07 -0700934extern int ceph_open(struct inode *inode, struct file *file);
Sage Weil5ef50c32012-07-31 11:27:36 -0700935extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
936 struct file *file, unsigned flags, umode_t mode,
937 int *opened);
Sage Weilde576062009-10-06 11:31:07 -0700938extern int ceph_release(struct inode *inode, struct file *filp);
Yan, Zheng31c542a2014-11-14 21:41:55 +0800939extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
940 char *data, size_t len);
Sage Weilde576062009-10-06 11:31:07 -0700941/* dir.c */
942extern const struct file_operations ceph_dir_fops;
Yan, Zheng38c48b52015-01-14 13:46:04 +0800943extern const struct file_operations ceph_snapdir_fops;
Sage Weilde576062009-10-06 11:31:07 -0700944extern const struct inode_operations ceph_dir_iops;
Yan, Zheng38c48b52015-01-14 13:46:04 +0800945extern const struct inode_operations ceph_snapdir_iops;
Sage Weil52dfb8a2010-08-03 10:25:30 -0700946extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
Sage Weilde576062009-10-06 11:31:07 -0700947 ceph_snapdir_dentry_ops;
948
949extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
Sage Weil468640e2011-07-26 11:28:11 -0700950extern int ceph_handle_snapdir(struct ceph_mds_request *req,
951 struct dentry *dentry, int err);
Sage Weilde576062009-10-06 11:31:07 -0700952extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
953 struct dentry *dentry, int err);
954
955extern void ceph_dentry_lru_add(struct dentry *dn);
956extern void ceph_dentry_lru_touch(struct dentry *dn);
957extern void ceph_dentry_lru_del(struct dentry *dn);
Sage Weil81a6cf22010-05-14 09:35:38 -0700958extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
Sage Weile5f86dc2011-07-26 11:30:55 -0700959extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800960extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
Sage Weilde576062009-10-06 11:31:07 -0700961
962/*
963 * our d_ops vary depending on whether the inode is live,
964 * snapshotted (read-only), or a virtual ".snap" directory.
965 */
966int ceph_init_dentry(struct dentry *dentry);
967
968
969/* ioctl.c */
970extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
971
972/* export.c */
973extern const struct export_operations ceph_export_ops;
974
Greg Farnum40819f62010-08-02 15:34:23 -0700975/* locks.c */
Yan, Zhengeb13e832014-03-09 23:16:40 +0800976extern __init void ceph_flock_init(void);
Greg Farnum40819f62010-08-02 15:34:23 -0700977extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
978extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
979extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
Jim Schutt39be95e2013-05-15 13:03:35 -0500980extern int ceph_encode_locks_to_buffer(struct inode *inode,
981 struct ceph_filelock *flocks,
982 int num_fcntl_locks,
983 int num_flock_locks);
984extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
985 struct ceph_pagelist *pagelist,
986 int num_fcntl_locks, int num_flock_locks);
Greg Farnum40819f62010-08-02 15:34:23 -0700987extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c);
988
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700989/* debugfs.c */
990extern int ceph_fs_debugfs_init(struct ceph_fs_client *client);
991extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
992
Sage Weilde576062009-10-06 11:31:07 -0700993#endif /* _FS_CEPH_SUPER_H */