blob: 0168b49fb6adb2be232a4861005c8f290f452701 [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) */
Yan, Zheng235a0982016-03-30 17:18:34 +080065 int mds_namespace;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070066
67 /*
68 * everything above this point can be memcmp'd; everything below
69 * is handled in compare_mount_options()
70 */
71
Sage Weilde576062009-10-06 11:31:07 -070072 char *snapdir_name; /* default ".snap" */
Yan, Zheng3f384952016-04-21 11:09:55 +080073 char *server_path; /* default "/" */
Sage Weilde576062009-10-06 11:31:07 -070074};
75
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070076struct ceph_fs_client {
Sage Weilde576062009-10-06 11:31:07 -070077 struct super_block *sb;
78
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070079 struct ceph_mount_options *mount_options;
80 struct ceph_client *client;
81
Sage Weilde576062009-10-06 11:31:07 -070082 unsigned long mount_state;
Sage Weil85ccce42010-02-17 10:02:43 -080083 int min_caps; /* min caps i added */
84
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070085 struct ceph_mds_client *mdsc;
Sage Weilde576062009-10-06 11:31:07 -070086
87 /* writeback */
88 mempool_t *wb_pagevec_pool;
89 struct workqueue_struct *wb_wq;
90 struct workqueue_struct *pg_inv_wq;
91 struct workqueue_struct *trunc_wq;
Yehuda Sadeh2baba252009-12-18 13:51:57 -080092 atomic_long_t writeback_count;
Sage Weilde576062009-10-06 11:31:07 -070093
94 struct backing_dev_info backing_dev_info;
Sage Weil07433042009-11-18 16:50:41 -080095
96#ifdef CONFIG_DEBUG_FS
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070097 struct dentry *debugfs_dentry_lru, *debugfs_caps;
Yehuda Sadeh2baba252009-12-18 13:51:57 -080098 struct dentry *debugfs_congestion_kb;
Sage Weil06edf042009-12-15 14:44:32 -080099 struct dentry *debugfs_bdi;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700100 struct dentry *debugfs_mdsc, *debugfs_mdsmap;
John Spray14ed9702014-09-12 16:58:49 +0100101 struct dentry *debugfs_mds_sessions;
Sage Weil07433042009-11-18 16:50:41 -0800102#endif
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400103
104#ifdef CONFIG_CEPH_FSCACHE
105 struct fscache_cookie *fscache;
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400106#endif
Sage Weilde576062009-10-06 11:31:07 -0700107};
108
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700109
Sage Weilde576062009-10-06 11:31:07 -0700110/*
111 * File i/o capability. This tracks shared state with the metadata
112 * server that allows us to cache or writeback attributes or to read
113 * and write data. For any given inode, we should have one or more
114 * capabilities, one issued by each metadata server, and our
115 * cumulative access is the OR of all issued capabilities.
116 *
117 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
118 * session capability lists.
119 */
120struct ceph_cap {
121 struct ceph_inode_info *ci;
122 struct rb_node ci_node; /* per-ci cap tree */
123 struct ceph_mds_session *session;
124 struct list_head session_caps; /* per-session caplist */
Sage Weilde576062009-10-06 11:31:07 -0700125 u64 cap_id; /* unique cap id (mds provided) */
Yan, Zheng745a8e32015-05-14 17:22:42 +0800126 union {
127 /* in-use caps */
128 struct {
129 int issued; /* latest, from the mds */
130 int implemented; /* implemented superset of
131 issued (for revocation) */
132 int mds, mds_wanted;
133 };
134 /* caps to release */
135 struct {
136 u64 cap_ino;
137 int queue_release;
138 };
139 };
Sage Weil685f9a5d2009-11-09 12:05:48 -0800140 u32 seq, issue_seq, mseq;
141 u32 cap_gen; /* active/stale cycle */
Sage Weilde576062009-10-06 11:31:07 -0700142 unsigned long last_used;
143 struct list_head caps_item;
144};
145
146#define CHECK_CAPS_NODELAY 1 /* do not delay any further */
147#define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */
148#define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */
149
150/*
151 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
152 * we first complete any in-process sync writes and writeback any dirty
153 * data before flushing the snapped state (tracked here) back to the MDS.
154 */
155struct ceph_cap_snap {
156 atomic_t nref;
157 struct ceph_inode_info *ci;
158 struct list_head ci_item, flushing_item;
159
160 u64 follows, flush_tid;
161 int issued, dirty;
162 struct ceph_snap_context *context;
163
Al Viro5706b272011-07-26 04:52:22 -0400164 umode_t mode;
Eric W. Biederman05cb11c2013-01-31 02:56:19 -0800165 kuid_t uid;
166 kgid_t gid;
Sage Weilde576062009-10-06 11:31:07 -0700167
Sage Weil4a625be2010-08-22 15:03:56 -0700168 struct ceph_buffer *xattr_blob;
Sage Weilde576062009-10-06 11:31:07 -0700169 u64 xattr_version;
170
171 u64 size;
172 struct timespec mtime, atime, ctime;
173 u64 time_warp_seq;
174 int writing; /* a sync write is still in progress */
175 int dirty_pages; /* dirty pages awaiting writeback */
Yan, Zhenge20d2582014-11-14 22:39:13 +0800176 bool inline_data;
Yan, Zheng86056092015-05-01 16:57:16 +0800177 bool need_flush;
Sage Weilde576062009-10-06 11:31:07 -0700178};
179
180static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
181{
Sage Weil4a625be2010-08-22 15:03:56 -0700182 if (atomic_dec_and_test(&capsnap->nref)) {
183 if (capsnap->xattr_blob)
184 ceph_buffer_put(capsnap->xattr_blob);
Sage Weilde576062009-10-06 11:31:07 -0700185 kfree(capsnap);
Sage Weil4a625be2010-08-22 15:03:56 -0700186 }
Sage Weilde576062009-10-06 11:31:07 -0700187}
188
Yan, Zheng553adfd2015-06-09 15:48:57 +0800189struct ceph_cap_flush {
190 u64 tid;
191 int caps;
Yan, Zhenge548e9b2015-06-10 15:17:56 +0800192 struct rb_node g_node; // global
Yan, Zheng553adfd2015-06-09 15:48:57 +0800193 union {
Yan, Zhenge548e9b2015-06-10 15:17:56 +0800194 struct rb_node i_node; // inode
Yan, Zheng553adfd2015-06-09 15:48:57 +0800195 struct list_head list;
196 };
197};
198
Sage Weilde576062009-10-06 11:31:07 -0700199/*
200 * The frag tree describes how a directory is fragmented, potentially across
201 * multiple metadata servers. It is also used to indicate points where
202 * metadata authority is delegated, and whether/where metadata is replicated.
203 *
204 * A _leaf_ frag will be present in the i_fragtree IFF there is
205 * delegation info. That is, if mds >= 0 || ndist > 0.
206 */
207#define CEPH_MAX_DIRFRAG_REP 4
208
209struct ceph_inode_frag {
210 struct rb_node node;
211
212 /* fragtree state */
213 u32 frag;
214 int split_by; /* i.e. 2^(split_by) children */
215
216 /* delegation and replication info */
217 int mds; /* -1 if same authority as parent */
218 int ndist; /* >0 if replicated */
219 int dist[CEPH_MAX_DIRFRAG_REP];
220};
221
222/*
223 * We cache inode xattrs as an encoded blob until they are first used,
224 * at which point we parse them into an rbtree.
225 */
226struct ceph_inode_xattr {
227 struct rb_node node;
228
229 const char *name;
230 int name_len;
231 const char *val;
232 int val_len;
233 int dirty;
234
235 int should_free_name;
236 int should_free_val;
237};
238
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700239/*
240 * Ceph dentry state
241 */
242struct ceph_dentry_info {
243 struct ceph_mds_session *lease_session;
244 u32 lease_gen, lease_shared_gen;
245 u32 lease_seq;
246 unsigned long lease_renew_after, lease_renew_from;
247 struct list_head lru;
248 struct dentry *dentry;
249 u64 time;
250 u64 offset;
251};
252
Sage Weilde576062009-10-06 11:31:07 -0700253struct ceph_inode_xattrs_info {
254 /*
255 * (still encoded) xattr blob. we avoid the overhead of parsing
256 * this until someone actually calls getxattr, etc.
257 *
258 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
259 * NULL means we don't know.
260 */
261 struct ceph_buffer *blob, *prealloc_blob;
262
263 struct rb_root index;
264 bool dirty;
265 int count;
266 int names_size;
267 int vals_size;
268 u64 version, index_version;
269};
270
271/*
272 * Ceph inode.
273 */
Sage Weilde576062009-10-06 11:31:07 -0700274struct ceph_inode_info {
275 struct ceph_vino i_vino; /* ceph ino + snap */
276
Sage Weilbe655592011-11-30 09:47:09 -0800277 spinlock_t i_ceph_lock;
278
Sage Weilde576062009-10-06 11:31:07 -0700279 u64 i_version;
Yan, Zheng31c542a2014-11-14 21:41:55 +0800280 u64 i_inline_version;
Sage Weilde576062009-10-06 11:31:07 -0700281 u32 i_time_warp_seq;
282
283 unsigned i_ceph_flags;
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800284 atomic64_t i_release_count;
285 atomic64_t i_ordered_count;
286 atomic64_t i_complete_seq[2];
Sage Weilde576062009-10-06 11:31:07 -0700287
Sage Weil6c0f3af2010-11-16 11:14:34 -0800288 struct ceph_dir_layout i_dir_layout;
Sage Weilde576062009-10-06 11:31:07 -0700289 struct ceph_file_layout i_layout;
Yan, Zheng5ea5c5e2016-02-14 18:06:41 +0800290 size_t i_pool_ns_len;
Sage Weilde576062009-10-06 11:31:07 -0700291 char *i_symlink;
292
293 /* for dirs */
294 struct timespec i_rctime;
295 u64 i_rbytes, i_rfiles, i_rsubdirs;
296 u64 i_files, i_subdirs;
Sage Weilde576062009-10-06 11:31:07 -0700297
298 struct rb_root i_fragtree;
Yan, Zheng1b1bc162016-05-04 11:40:30 +0800299 int i_fragtree_nsplits;
Sage Weilde576062009-10-06 11:31:07 -0700300 struct mutex i_fragtree_mutex;
301
302 struct ceph_inode_xattrs_info i_xattrs;
303
Sage Weilbe655592011-11-30 09:47:09 -0800304 /* capabilities. protected _both_ by i_ceph_lock and cap->session's
Sage Weilde576062009-10-06 11:31:07 -0700305 * s_mutex. */
306 struct rb_root i_caps; /* cap list */
307 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
308 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
309 struct list_head i_dirty_item, i_flushing_item;
Sage Weilde576062009-10-06 11:31:07 -0700310 /* we need to track cap writeback on a per-cap-bit basis, to allow
311 * overlapping, pipelined cap flushes to the mds. we can probably
312 * reduce the tid to 8 bits if we're concerned about inode size. */
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800313 struct ceph_cap_flush *i_prealloc_cap_flush;
Yan, Zheng553adfd2015-06-09 15:48:57 +0800314 struct rb_root i_cap_flush_tree;
Sage Weilde576062009-10-06 11:31:07 -0700315 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
316 unsigned long i_hold_caps_min; /* jiffies */
317 unsigned long i_hold_caps_max; /* jiffies */
318 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
Sage Weilde576062009-10-06 11:31:07 -0700319 struct ceph_cap_reservation i_cap_migration_resv;
320 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
Sage Weil7d8cb262010-08-24 08:44:16 -0700321 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or
322 dirty|flushing caps */
Sage Weilde576062009-10-06 11:31:07 -0700323 unsigned i_snap_caps; /* cap bits for snapped files */
324
325 int i_nr_by_mode[CEPH_FILE_MODE_NUM]; /* open file counts */
326
Yan, Zhengb0d7c222013-08-12 21:42:15 -0700327 struct mutex i_truncate_mutex;
Sage Weilde576062009-10-06 11:31:07 -0700328 u32 i_truncate_seq; /* last truncate to smaller size */
329 u64 i_truncate_size; /* and the size we last truncated down to */
330 int i_truncate_pending; /* still need to call vmtruncate */
331
332 u64 i_max_size; /* max file size authorized by mds */
333 u64 i_reported_size; /* (max_)size reported to or requested of mds */
334 u64 i_wanted_max_size; /* offset we'd like to write too */
335 u64 i_requested_max_size; /* max_size we've requested */
336
337 /* held references to caps */
338 int i_pin_ref;
Henry C Changd3d07202011-05-11 10:29:54 +0000339 int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref;
Sage Weilde576062009-10-06 11:31:07 -0700340 int i_wrbuffer_ref, i_wrbuffer_ref_head;
341 u32 i_shared_gen; /* increment each time we get FILE_SHARED */
Sage Weilcd045cb2010-11-04 11:05:05 -0700342 u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */
Sage Weilde576062009-10-06 11:31:07 -0700343 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
344
345 struct list_head i_unsafe_writes; /* uncommitted sync writes */
346 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
Yan, Zheng68cd5b42015-10-27 18:36:06 +0800347 struct list_head i_unsafe_iops; /* uncommitted mds inode ops */
Sage Weilde576062009-10-06 11:31:07 -0700348 spinlock_t i_unsafe_lock;
349
350 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
351 int i_snap_realm_counter; /* snap realm (if caps) */
352 struct list_head i_snap_realm_item;
353 struct list_head i_snap_flush_item;
354
355 struct work_struct i_wb_work; /* writeback work */
356 struct work_struct i_pg_inv_work; /* page invalidation work */
357
358 struct work_struct i_vmtruncate_work;
359
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400360#ifdef CONFIG_CEPH_FSCACHE
361 struct fscache_cookie *fscache;
Yan, Zhengf7f7e7a2016-05-18 20:31:55 +0800362 u32 i_fscache_gen;
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400363#endif
Sage Weilde576062009-10-06 11:31:07 -0700364 struct inode vfs_inode; /* at end */
365};
366
367static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
368{
Noah Watkinsfbbccec2009-10-28 11:54:49 -0700369 return container_of(inode, struct ceph_inode_info, vfs_inode);
Sage Weilde576062009-10-06 11:31:07 -0700370}
371
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800372static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode)
373{
374 return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
375}
376
377static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb)
378{
379 return (struct ceph_fs_client *)sb->s_fs_info;
380}
381
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700382static inline struct ceph_vino ceph_vino(struct inode *inode)
383{
384 return ceph_inode(inode)->i_vino;
385}
386
387/*
388 * ino_t is <64 bits on many architectures, blech.
389 *
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800390 * i_ino (kernel inode) st_ino (userspace)
391 * i386 32 32
392 * x86_64+ino32 64 32
393 * x86_64 64 64
394 */
Amon Ott3310f752011-10-20 13:04:07 -0700395static inline u32 ceph_ino_to_ino32(__u64 vino)
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800396{
Amon Ott3310f752011-10-20 13:04:07 -0700397 u32 ino = vino & 0xffffffff;
398 ino ^= vino >> 32;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800399 if (!ino)
Amon Otta661fc52012-01-23 09:25:23 -0800400 ino = 2;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800401 return ino;
402}
403
404/*
405 * kernel i_ino value
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700406 */
407static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
408{
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700409#if BITS_PER_LONG == 32
Amon Ott3310f752011-10-20 13:04:07 -0700410 return ceph_ino_to_ino32(vino.ino);
411#else
412 return (ino_t)vino.ino;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700413#endif
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700414}
415
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800416/*
417 * user-visible ino (stat, filldir)
418 */
419#if BITS_PER_LONG == 32
420static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
421{
422 return ino;
423}
424#else
425static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
426{
427 if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))
428 ino = ceph_ino_to_ino32(ino);
429 return ino;
430}
431#endif
432
433
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700434/* for printf-style formatting */
435#define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
436
437static inline u64 ceph_ino(struct inode *inode)
438{
439 return ceph_inode(inode)->i_vino.ino;
440}
441static inline u64 ceph_snap(struct inode *inode)
442{
443 return ceph_inode(inode)->i_vino.snap;
444}
445
446static inline int ceph_ino_compare(struct inode *inode, void *data)
447{
448 struct ceph_vino *pvino = (struct ceph_vino *)data;
449 struct ceph_inode_info *ci = ceph_inode(inode);
450 return ci->i_vino.ino == pvino->ino &&
451 ci->i_vino.snap == pvino->snap;
452}
453
454static inline struct inode *ceph_find_inode(struct super_block *sb,
455 struct ceph_vino vino)
456{
457 ino_t t = ceph_vino_to_ino(vino);
458 return ilookup5(sb, t, ceph_ino_compare, &vino);
459}
460
461
462/*
463 * Ceph inode.
464 */
Yan, Zheng10183a62015-04-27 15:33:28 +0800465#define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */
466#define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */
467#define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */
468#define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */
469#define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */
470#define CEPH_I_POOL_RD (1 << 5) /* can read from pool */
471#define CEPH_I_POOL_WR (1 << 6) /* can write to pool */
Yan, Zheng315f2402016-03-07 10:34:50 +0800472#define CEPH_I_SEC_INITED (1 << 7) /* security initialized */
Yan, Zheng77310322016-04-08 15:27:16 +0800473#define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700474
Yan, Zheng2f276c52013-03-13 19:44:32 +0800475static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800476 long long release_count,
477 long long ordered_count)
Sage Weilde576062009-10-06 11:31:07 -0700478{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800479 smp_mb__before_atomic();
480 atomic64_set(&ci->i_complete_seq[0], release_count);
481 atomic64_set(&ci->i_complete_seq[1], ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700482}
483
Yan, Zheng2f276c52013-03-13 19:44:32 +0800484static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700485{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800486 atomic64_inc(&ci->i_release_count);
487}
488
489static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
490{
491 atomic64_inc(&ci->i_ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700492}
493
Yan, Zheng2f276c52013-03-13 19:44:32 +0800494static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700495{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800496 return atomic64_read(&ci->i_complete_seq[0]) ==
497 atomic64_read(&ci->i_release_count);
Yan, Zheng2f276c52013-03-13 19:44:32 +0800498}
Sage Weilde576062009-10-06 11:31:07 -0700499
Yan, Zheng70db4f32014-10-21 18:09:56 -0700500static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
501{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800502 return atomic64_read(&ci->i_complete_seq[0]) ==
503 atomic64_read(&ci->i_release_count) &&
504 atomic64_read(&ci->i_complete_seq[1]) ==
505 atomic64_read(&ci->i_ordered_count);
Yan, Zheng70db4f32014-10-21 18:09:56 -0700506}
507
Yan, Zheng2f276c52013-03-13 19:44:32 +0800508static inline void ceph_dir_clear_complete(struct inode *inode)
509{
510 __ceph_dir_clear_complete(ceph_inode(inode));
511}
512
Yan, Zheng70db4f32014-10-21 18:09:56 -0700513static inline void ceph_dir_clear_ordered(struct inode *inode)
Yan, Zheng2f276c52013-03-13 19:44:32 +0800514{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800515 __ceph_dir_clear_ordered(ceph_inode(inode));
Sage Weilde576062009-10-06 11:31:07 -0700516}
517
Yan, Zheng70db4f32014-10-21 18:09:56 -0700518static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
519{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800520 bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
521 smp_rmb();
Yan, Zheng70db4f32014-10-21 18:09:56 -0700522 return ret;
523}
Sage Weilde576062009-10-06 11:31:07 -0700524
525/* find a specific frag @f */
526extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
527 u32 f);
528
529/*
530 * choose fragment for value @v. copy frag content to pfrag, if leaf
531 * exists
532 */
533extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
534 struct ceph_inode_frag *pfrag,
535 int *found);
536
Sage Weilde576062009-10-06 11:31:07 -0700537static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
538{
539 return (struct ceph_dentry_info *)dentry->d_fsdata;
540}
541
Sage Weilde576062009-10-06 11:31:07 -0700542/*
543 * caps helpers
544 */
545static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
546{
547 return !RB_EMPTY_ROOT(&ci->i_caps);
548}
549
550extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
551extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
552extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
553 struct ceph_cap *cap);
554
555static inline int ceph_caps_issued(struct ceph_inode_info *ci)
556{
557 int issued;
Sage Weilbe655592011-11-30 09:47:09 -0800558 spin_lock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700559 issued = __ceph_caps_issued(ci, NULL);
Sage Weilbe655592011-11-30 09:47:09 -0800560 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700561 return issued;
562}
563
564static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
565 int touch)
566{
567 int r;
Sage Weilbe655592011-11-30 09:47:09 -0800568 spin_lock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700569 r = __ceph_caps_issued_mask(ci, mask, touch);
Sage Weilbe655592011-11-30 09:47:09 -0800570 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700571 return r;
572}
573
574static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
575{
576 return ci->i_dirty_caps | ci->i_flushing_caps;
577}
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800578extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
579extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
580extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
581 struct ceph_cap_flush **pcf);
Sage Weilde576062009-10-06 11:31:07 -0700582
Yan, Zheng9563f882013-11-22 13:50:45 +0800583extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
584 struct ceph_cap *ocap, int mask);
Sage Weilde576062009-10-06 11:31:07 -0700585extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
586extern int __ceph_caps_used(struct ceph_inode_info *ci);
587
588extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
589
590/*
591 * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
592 */
593static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
594{
595 int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
596 if (w & CEPH_CAP_FILE_BUFFER)
597 w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */
598 return w;
599}
600
601/* what the mds thinks we want */
602extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
603
Yehuda Sadeh37151662010-06-17 16:16:12 -0700604extern void ceph_caps_init(struct ceph_mds_client *mdsc);
605extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
606extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
majianpeng93faca62013-06-26 11:15:27 +0800607extern void ceph_reserve_caps(struct ceph_mds_client *mdsc,
Yehuda Sadeh37151662010-06-17 16:16:12 -0700608 struct ceph_cap_reservation *ctx, int need);
609extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
610 struct ceph_cap_reservation *ctx);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700611extern void ceph_reservation_status(struct ceph_fs_client *client,
Sage Weilde576062009-10-06 11:31:07 -0700612 int *total, int *avail, int *used,
Sage Weil85ccce42010-02-17 10:02:43 -0800613 int *reserved, int *min);
Sage Weilde576062009-10-06 11:31:07 -0700614
Sage Weilde576062009-10-06 11:31:07 -0700615
Sage Weilde576062009-10-06 11:31:07 -0700616
617/*
618 * we keep buffered readdir results attached to file->private_data
619 */
Sage Weil4918b6d2011-07-26 11:26:07 -0700620#define CEPH_F_SYNC 1
Sage Weil9cfa1092011-07-26 11:26:18 -0700621#define CEPH_F_ATEND 2
Sage Weil4918b6d2011-07-26 11:26:07 -0700622
Sage Weilde576062009-10-06 11:31:07 -0700623struct ceph_file_info {
Sage Weil252c6722011-07-26 11:25:27 -0700624 short fmode; /* initialized on open */
625 short flags; /* CEPH_F_* */
Sage Weilde576062009-10-06 11:31:07 -0700626
627 /* readdir: position within the dir */
628 u32 frag;
629 struct ceph_mds_request *last_readdir;
Sage Weilde576062009-10-06 11:31:07 -0700630
631 /* readdir: position within a frag */
Yan, Zhengf0494202014-02-27 16:26:24 +0800632 unsigned next_offset; /* offset of next chunk (last_name's + 1) */
Sage Weilde576062009-10-06 11:31:07 -0700633 char *last_name; /* last entry in previous chunk */
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800634 long long dir_release_count;
635 long long dir_ordered_count;
636 int readdir_cache_idx;
Sage Weilde576062009-10-06 11:31:07 -0700637
638 /* used for -o dirstat read() on directory thing */
639 char *dir_info;
640 int dir_info_len;
641};
642
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800643struct ceph_readdir_cache_control {
644 struct page *page;
645 struct dentry **dentries;
646 int index;
647};
Sage Weilde576062009-10-06 11:31:07 -0700648
649/*
Sage Weilde576062009-10-06 11:31:07 -0700650 * A "snap realm" describes a subset of the file hierarchy sharing
651 * the same set of snapshots that apply to it. The realms themselves
652 * are organized into a hierarchy, such that children inherit (some of)
653 * the snapshots of their parents.
654 *
655 * All inodes within the realm that have capabilities are linked into a
656 * per-realm list.
657 */
658struct ceph_snap_realm {
659 u64 ino;
660 atomic_t nref;
Sage Weila105f002010-02-15 14:37:55 -0800661 struct rb_node node;
662
Sage Weilde576062009-10-06 11:31:07 -0700663 u64 created, seq;
664 u64 parent_ino;
665 u64 parent_since; /* snapid when our current parent became so */
666
667 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
Alex Elderaa711ee2012-07-13 20:35:11 -0500668 u32 num_prior_parent_snaps; /* had prior to parent_since */
Sage Weilde576062009-10-06 11:31:07 -0700669 u64 *snaps; /* snaps specific to this realm */
Alex Elderaa711ee2012-07-13 20:35:11 -0500670 u32 num_snaps;
Sage Weilde576062009-10-06 11:31:07 -0700671
672 struct ceph_snap_realm *parent;
673 struct list_head children; /* list of child realms */
674 struct list_head child_item;
675
676 struct list_head empty_item; /* if i have ref==0 */
677
Sage Weilae00d4f2010-09-16 16:26:51 -0700678 struct list_head dirty_item; /* if realm needs new context */
679
Sage Weilde576062009-10-06 11:31:07 -0700680 /* the current set of snaps for this realm */
681 struct ceph_snap_context *cached_context;
682
683 struct list_head inodes_with_caps;
684 spinlock_t inodes_with_caps_lock;
685};
686
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700687static inline int default_congestion_kb(void)
Sage Weilde576062009-10-06 11:31:07 -0700688{
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700689 int congestion_kb;
690
691 /*
692 * Copied from NFS
693 *
694 * congestion size, scale with available memory.
695 *
696 * 64MB: 8192k
697 * 128MB: 11585k
698 * 256MB: 16384k
699 * 512MB: 23170k
700 * 1GB: 32768k
701 * 2GB: 46340k
702 * 4GB: 65536k
703 * 8GB: 92681k
704 * 16GB: 131072k
705 *
706 * This allows larger machines to have larger/more transfers.
707 * Limit the default to 256M
708 */
709 congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
710 if (congestion_kb > 256*1024)
711 congestion_kb = 256*1024;
712
713 return congestion_kb;
Sage Weilde576062009-10-06 11:31:07 -0700714}
715
716
717
718/* snap.c */
719struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
720 u64 ino);
721extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
722 struct ceph_snap_realm *realm);
723extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
724 struct ceph_snap_realm *realm);
725extern int ceph_update_snap_trace(struct ceph_mds_client *m,
Yan, Zheng982d6012014-12-23 15:30:54 +0800726 void *p, void *e, bool deletion,
727 struct ceph_snap_realm **realm_ret);
Sage Weilde576062009-10-06 11:31:07 -0700728extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
Sage Weil2600d2d2010-02-22 15:12:16 -0800729 struct ceph_mds_session *session,
Sage Weilde576062009-10-06 11:31:07 -0700730 struct ceph_msg *msg);
Sage Weilfc837c8f2010-04-13 11:41:22 -0700731extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
Sage Weilde576062009-10-06 11:31:07 -0700732extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
733 struct ceph_cap_snap *capsnap);
734extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
735
736/*
737 * a cap_snap is "pending" if it is still awaiting an in-progress
738 * sync write (that may/may not still update size, mtime, etc.).
739 */
740static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
741{
742 return !list_empty(&ci->i_cap_snaps) &&
Yan, Zheng86056092015-05-01 16:57:16 +0800743 list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
744 ci_item)->writing;
Sage Weilde576062009-10-06 11:31:07 -0700745}
746
Sage Weilde576062009-10-06 11:31:07 -0700747/* inode.c */
748extern const struct inode_operations ceph_file_iops;
749
750extern struct inode *ceph_alloc_inode(struct super_block *sb);
751extern void ceph_destroy_inode(struct inode *inode);
Yan, Zheng9f12bd12013-09-20 19:55:31 +0800752extern int ceph_drop_inode(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700753
754extern struct inode *ceph_get_inode(struct super_block *sb,
755 struct ceph_vino vino);
756extern struct inode *ceph_get_snapdir(struct inode *parent);
757extern int ceph_fill_file_size(struct inode *inode, int issued,
758 u32 truncate_seq, u64 truncate_size, u64 size);
759extern void ceph_fill_file_time(struct inode *inode, int issued,
760 u64 time_warp_seq, struct timespec *ctime,
761 struct timespec *mtime, struct timespec *atime);
762extern int ceph_fill_trace(struct super_block *sb,
763 struct ceph_mds_request *req,
764 struct ceph_mds_session *session);
765extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
766 struct ceph_mds_session *session);
767
768extern int ceph_inode_holds_cap(struct inode *inode, int mask);
769
770extern int ceph_inode_set_size(struct inode *inode, loff_t size);
Yan, Zhengb415bf42013-07-02 12:40:19 +0800771extern void __ceph_do_pending_vmtruncate(struct inode *inode);
Sage Weil3c6f6b72010-02-09 15:24:44 -0800772extern void ceph_queue_vmtruncate(struct inode *inode);
773
774extern void ceph_queue_invalidate(struct inode *inode);
775extern void ceph_queue_writeback(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700776
Yan, Zheng01deead2014-11-14 21:56:29 +0800777extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
778 int mask, bool force);
779static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
780{
781 return __ceph_do_getattr(inode, NULL, mask, force);
782}
Al Viro10556cb2011-06-20 19:28:19 -0400783extern int ceph_permission(struct inode *inode, int mask);
Andreas Gruenbachera26fecc2016-04-14 00:30:16 +0200784extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
Sage Weilde576062009-10-06 11:31:07 -0700785extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
786extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
787 struct kstat *stat);
788
789/* xattr.c */
Andreas Gruenbachera26fecc2016-04-14 00:30:16 +0200790int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800791ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
Sage Weilde576062009-10-06 11:31:07 -0700792extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
Sage Weilde576062009-10-06 11:31:07 -0700793extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
794extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
Alex Elder3ce6cd12012-01-23 15:49:28 -0600795extern void __init ceph_xattr_init(void);
796extern void ceph_xattr_exit(void);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800797extern const struct xattr_handler *ceph_xattr_handlers[];
Sage Weilde576062009-10-06 11:31:07 -0700798
Yan, Zheng315f2402016-03-07 10:34:50 +0800799#ifdef CONFIG_SECURITY
800extern bool ceph_security_xattr_deadlock(struct inode *in);
801extern bool ceph_security_xattr_wanted(struct inode *in);
802#else
803static inline bool ceph_security_xattr_deadlock(struct inode *in)
804{
805 return false;
806}
807static inline bool ceph_security_xattr_wanted(struct inode *in)
808{
809 return false;
810}
811#endif
812
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800813/* acl.c */
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800814struct ceph_acls_info {
815 void *default_acl;
816 void *acl;
817 struct ceph_pagelist *pagelist;
818};
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800819
820#ifdef CONFIG_CEPH_FS_POSIX_ACL
821
822struct posix_acl *ceph_get_acl(struct inode *, int);
Sage Weil72466d02014-01-29 06:22:25 -0800823int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800824int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
825 struct ceph_acls_info *info);
826void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info);
827void ceph_release_acls_info(struct ceph_acls_info *info);
Guangliang Zhaoc969d9b2014-02-16 08:35:52 -0800828
829static inline void ceph_forget_all_cached_acls(struct inode *inode)
830{
831 forget_all_cached_acls(inode);
832}
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800833
834#else
835
836#define ceph_get_acl NULL
Sage Weil72466d02014-01-29 06:22:25 -0800837#define ceph_set_acl NULL
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800838
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800839static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
840 struct ceph_acls_info *info)
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800841{
842 return 0;
843}
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800844static inline void ceph_init_inode_acls(struct inode *inode,
845 struct ceph_acls_info *info)
846{
847}
848static inline void ceph_release_acls_info(struct ceph_acls_info *info)
849{
850}
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800851static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
852{
853 return 0;
854}
855
856static inline void ceph_forget_all_cached_acls(struct inode *inode)
857{
858}
859
860#endif
861
Sage Weilde576062009-10-06 11:31:07 -0700862/* caps.c */
863extern const char *ceph_cap_string(int c);
864extern void ceph_handle_caps(struct ceph_mds_session *session,
865 struct ceph_msg *msg);
Yan, Zhengd9df2782014-04-18 09:57:11 +0800866extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
867 struct ceph_cap_reservation *ctx);
868extern void ceph_add_cap(struct inode *inode,
869 struct ceph_mds_session *session, u64 cap_id,
870 int fmode, unsigned issued, unsigned wanted,
871 unsigned cap, unsigned seq, u64 realmino, int flags,
872 struct ceph_cap **new_cap);
Yan, Zhenga096b092013-09-22 10:15:58 +0800873extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
Yehuda Sadeh37151662010-06-17 16:16:12 -0700874extern void ceph_put_cap(struct ceph_mds_client *mdsc,
875 struct ceph_cap *cap);
Yan, Zheng9215aee2013-11-30 12:47:41 +0800876extern int ceph_is_any_caps(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700877
878extern void ceph_queue_caps_release(struct inode *inode);
Stephen Rothwellf1a3d572010-01-18 11:53:08 +1100879extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
Josef Bacik02c24a82011-07-16 20:44:56 -0400880extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
881 int datasync);
Yan, Zhenge548e9b2015-06-10 15:17:56 +0800882extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
883 struct ceph_mds_session *session);
Sage Weilde576062009-10-06 11:31:07 -0700884extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
885 struct ceph_mds_session *session);
Greg Farnum2bc50252010-06-30 12:44:34 -0700886extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
887 int mds);
Sage Weilde576062009-10-06 11:31:07 -0700888extern int ceph_get_cap_mds(struct inode *inode);
889extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
890extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
891extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
892 struct ceph_snap_context *snapc);
893extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
Sage Weile8351242010-09-17 08:03:08 -0700894 struct ceph_mds_session **psession,
895 int again);
Sage Weilde576062009-10-06 11:31:07 -0700896extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
897 struct ceph_mds_session *session);
Sage Weilafcdaea2009-10-14 14:27:38 -0700898extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
899extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
Sage Weilde576062009-10-06 11:31:07 -0700900
901extern int ceph_encode_inode_release(void **p, struct inode *inode,
902 int mds, int drop, int unless, int force);
903extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
904 int mds, int drop, int unless);
905
906extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
Yan, Zheng31c542a2014-11-14 21:41:55 +0800907 loff_t endoff, int *got, struct page **pinned_page);
Sage Weilde576062009-10-06 11:31:07 -0700908
909/* for counting open files by mode */
910static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
911{
912 ci->i_nr_by_mode[mode]++;
913}
914extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
915
916/* addr.c */
917extern const struct address_space_operations ceph_aops;
918extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
Yan, Zheng10183a62015-04-27 15:33:28 +0800919extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
920extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need);
921extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
Sage Weilde576062009-10-06 11:31:07 -0700922
923/* file.c */
924extern const struct file_operations ceph_file_fops;
Alex Elder9e0eb852013-02-06 13:11:38 -0600925
Yan, Zheng77310322016-04-08 15:27:16 +0800926extern int ceph_renew_caps(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700927extern int ceph_open(struct inode *inode, struct file *file);
Sage Weil5ef50c32012-07-31 11:27:36 -0700928extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
929 struct file *file, unsigned flags, umode_t mode,
930 int *opened);
Sage Weilde576062009-10-06 11:31:07 -0700931extern int ceph_release(struct inode *inode, struct file *filp);
Yan, Zheng31c542a2014-11-14 21:41:55 +0800932extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
933 char *data, size_t len);
Sage Weilde576062009-10-06 11:31:07 -0700934/* dir.c */
935extern const struct file_operations ceph_dir_fops;
Yan, Zheng38c48b52015-01-14 13:46:04 +0800936extern const struct file_operations ceph_snapdir_fops;
Sage Weilde576062009-10-06 11:31:07 -0700937extern const struct inode_operations ceph_dir_iops;
Yan, Zheng38c48b52015-01-14 13:46:04 +0800938extern const struct inode_operations ceph_snapdir_iops;
Sage Weil52dfb8a2010-08-03 10:25:30 -0700939extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
Sage Weilde576062009-10-06 11:31:07 -0700940 ceph_snapdir_dentry_ops;
941
Yan, Zhengf3c4ebe2016-04-29 11:27:30 +0800942extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
Sage Weilde576062009-10-06 11:31:07 -0700943extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
Sage Weil468640e2011-07-26 11:28:11 -0700944extern int ceph_handle_snapdir(struct ceph_mds_request *req,
945 struct dentry *dentry, int err);
Sage Weilde576062009-10-06 11:31:07 -0700946extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
947 struct dentry *dentry, int err);
948
949extern void ceph_dentry_lru_add(struct dentry *dn);
950extern void ceph_dentry_lru_touch(struct dentry *dn);
951extern void ceph_dentry_lru_del(struct dentry *dn);
Sage Weil81a6cf22010-05-14 09:35:38 -0700952extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
Sage Weile5f86dc2011-07-26 11:30:55 -0700953extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800954extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
Sage Weilde576062009-10-06 11:31:07 -0700955
956/*
957 * our d_ops vary depending on whether the inode is live,
958 * snapshotted (read-only), or a virtual ".snap" directory.
959 */
960int ceph_init_dentry(struct dentry *dentry);
961
962
963/* ioctl.c */
964extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
965
966/* export.c */
967extern const struct export_operations ceph_export_ops;
968
Greg Farnum40819f62010-08-02 15:34:23 -0700969/* locks.c */
Yan, Zhengeb13e832014-03-09 23:16:40 +0800970extern __init void ceph_flock_init(void);
Greg Farnum40819f62010-08-02 15:34:23 -0700971extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
972extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
973extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
Jim Schutt39be95e2013-05-15 13:03:35 -0500974extern int ceph_encode_locks_to_buffer(struct inode *inode,
975 struct ceph_filelock *flocks,
976 int num_fcntl_locks,
977 int num_flock_locks);
978extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
979 struct ceph_pagelist *pagelist,
980 int num_fcntl_locks, int num_flock_locks);
Greg Farnum40819f62010-08-02 15:34:23 -0700981extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c);
982
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700983/* debugfs.c */
984extern int ceph_fs_debugfs_init(struct ceph_fs_client *client);
985extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
986
Sage Weilde576062009-10-06 11:31:07 -0700987#endif /* _FS_CEPH_SUPER_H */