1 #ifndef _FS_CEPH_SUPER_H
2 #define _FS_CEPH_SUPER_H
4 #include "ceph_debug.h"
6 #include <asm/unaligned.h>
7 #include <linux/backing-dev.h>
8 #include <linux/completion.h>
9 #include <linux/exportfs.h>
11 #include <linux/mempool.h>
12 #include <linux/pagemap.h>
13 #include <linux/wait.h>
16 #include "messenger.h"
18 #include "mon_client.h"
19 #include "mds_client.h"
20 #include "osd_client.h"
23 /* f_type in struct statfs */
24 #define CEPH_SUPER_MAGIC 0x00c36400
26 /* large granularity for statfs utilization stats to facilitate
27 * large volume sizes on 32-bit machines. */
28 #define CEPH_BLOCK_SHIFT 20 /* 1 MB */
29 #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT)
34 #define CEPH_OPT_FSID (1<<0)
35 #define CEPH_OPT_NOSHARE (1<<1) /* don't share client with other sbs */
36 #define CEPH_OPT_MYIP (1<<2) /* specified my ip */
37 #define CEPH_OPT_DIRSTAT (1<<4) /* funky `cat dirname` for stats */
38 #define CEPH_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */
39 #define CEPH_OPT_NOCRC (1<<6) /* no data crc on writes */
40 #define CEPH_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */
42 #define CEPH_OPT_DEFAULT (CEPH_OPT_RBYTES)
44 #define ceph_set_opt(client, opt) \
45 (client)->mount_args->flags |= CEPH_OPT_##opt;
46 #define ceph_test_opt(client, opt) \
47 (!!((client)->mount_args->flags & CEPH_OPT_##opt))
50 struct ceph_mount_args {
53 struct ceph_entity_addr *mon_addr;
56 int caps_wanted_delay_min, caps_wanted_delay_max;
57 struct ceph_fsid fsid;
58 struct ceph_entity_addr my_addr;
60 int rsize; /* max readahead */
61 int max_readdir; /* max readdir size */
63 char *snapdir_name; /* default ".snap" */
66 int cap_release_safety;
72 #define CEPH_MOUNT_TIMEOUT_DEFAULT 60
73 #define CEPH_MOUNT_RSIZE_DEFAULT (512*1024) /* readahead */
75 #define CEPH_MSG_MAX_FRONT_LEN (16*1024*1024)
76 #define CEPH_MSG_MAX_DATA_LEN (16*1024*1024)
78 #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
79 #define CEPH_AUTH_NAME_DEFAULT "guest"
82 * Delay telling the MDS we no longer want caps, in case we reopen
83 * the file. Delay a minimum amount of time, even if we send a cap
84 * message for some other reason. Otherwise, take the oppotunity to
85 * update the mds to avoid sending another message later.
87 #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */
88 #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */
95 CEPH_MOUNT_UNMOUNTING,
103 static inline unsigned long time_sub(unsigned long a, unsigned long b)
105 BUG_ON(time_after(b, a));
106 return (long)a - (long)b;
110 * per-filesystem client state
112 * possibly shared by multiple mount points, if they are
113 * mounting the same ceph filesystem/cluster.
116 struct ceph_fsid fsid;
119 struct mutex mount_mutex; /* serialize mount attempts */
120 struct ceph_mount_args *mount_args;
122 struct super_block *sb;
124 unsigned long mount_state;
125 wait_queue_head_t mount_wq;
129 struct ceph_messenger *msgr; /* messenger instance */
130 struct ceph_mon_client monc;
131 struct ceph_mds_client mdsc;
132 struct ceph_osd_client osdc;
135 mempool_t *wb_pagevec_pool;
136 struct workqueue_struct *wb_wq;
137 struct workqueue_struct *pg_inv_wq;
138 struct workqueue_struct *trunc_wq;
140 struct backing_dev_info backing_dev_info;
142 #ifdef CONFIG_DEBUG_FS
143 struct dentry *debugfs_monmap;
144 struct dentry *debugfs_mdsmap, *debugfs_osdmap;
145 struct dentry *debugfs_dir, *debugfs_dentry_lru, *debugfs_caps;
149 static inline struct ceph_client *ceph_client(struct super_block *sb)
151 return sb->s_fs_info;
156 * File i/o capability. This tracks shared state with the metadata
157 * server that allows us to cache or writeback attributes or to read
158 * and write data. For any given inode, we should have one or more
159 * capabilities, one issued by each metadata server, and our
160 * cumulative access is the OR of all issued capabilities.
162 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
163 * session capability lists.
166 struct ceph_inode_info *ci;
167 struct rb_node ci_node; /* per-ci cap tree */
168 struct ceph_mds_session *session;
169 struct list_head session_caps; /* per-session caplist */
171 u64 cap_id; /* unique cap id (mds provided) */
172 int issued; /* latest, from the mds */
173 int implemented; /* implemented superset of issued (for revocation) */
175 u32 seq, issue_seq, mseq;
176 u32 cap_gen; /* active/stale cycle */
177 unsigned long last_used;
178 struct list_head caps_item;
181 #define CHECK_CAPS_NODELAY 1 /* do not delay any further */
182 #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */
183 #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */
186 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
187 * we first complete any in-process sync writes and writeback any dirty
188 * data before flushing the snapped state (tracked here) back to the MDS.
190 struct ceph_cap_snap {
192 struct ceph_inode_info *ci;
193 struct list_head ci_item, flushing_item;
195 u64 follows, flush_tid;
197 struct ceph_snap_context *context;
208 struct timespec mtime, atime, ctime;
210 int writing; /* a sync write is still in progress */
211 int dirty_pages; /* dirty pages awaiting writeback */
214 static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
216 if (atomic_dec_and_test(&capsnap->nref))
221 * The frag tree describes how a directory is fragmented, potentially across
222 * multiple metadata servers. It is also used to indicate points where
223 * metadata authority is delegated, and whether/where metadata is replicated.
225 * A _leaf_ frag will be present in the i_fragtree IFF there is
226 * delegation info. That is, if mds >= 0 || ndist > 0.
228 #define CEPH_MAX_DIRFRAG_REP 4
230 struct ceph_inode_frag {
235 int split_by; /* i.e. 2^(split_by) children */
237 /* delegation and replication info */
238 int mds; /* -1 if same authority as parent */
239 int ndist; /* >0 if replicated */
240 int dist[CEPH_MAX_DIRFRAG_REP];
244 * We cache inode xattrs as an encoded blob until they are first used,
245 * at which point we parse them into an rbtree.
247 struct ceph_inode_xattr {
256 int should_free_name;
260 struct ceph_inode_xattrs_info {
262 * (still encoded) xattr blob. we avoid the overhead of parsing
263 * this until someone actually calls getxattr, etc.
265 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
266 * NULL means we don't know.
268 struct ceph_buffer *blob, *prealloc_blob;
270 struct rb_root index;
275 u64 version, index_version;
281 #define CEPH_I_COMPLETE 1 /* we have complete directory cached */
282 #define CEPH_I_NODELAY 4 /* do not delay cap release */
283 #define CEPH_I_FLUSH 8 /* do not delay flush of dirty metadata */
285 struct ceph_inode_info {
286 struct ceph_vino i_vino; /* ceph ino + snap */
291 unsigned i_ceph_flags;
292 unsigned long i_release_count;
294 struct ceph_file_layout i_layout;
298 struct timespec i_rctime;
299 u64 i_rbytes, i_rfiles, i_rsubdirs;
300 u64 i_files, i_subdirs;
301 u64 i_max_offset; /* largest readdir offset, set with I_COMPLETE */
303 struct rb_root i_fragtree;
304 struct mutex i_fragtree_mutex;
306 struct ceph_inode_xattrs_info i_xattrs;
308 /* capabilities. protected _both_ by i_lock and cap->session's
310 struct rb_root i_caps; /* cap list */
311 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
312 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
313 struct list_head i_dirty_item, i_flushing_item;
315 /* we need to track cap writeback on a per-cap-bit basis, to allow
316 * overlapping, pipelined cap flushes to the mds. we can probably
317 * reduce the tid to 8 bits if we're concerned about inode size. */
318 u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS];
319 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
320 unsigned long i_hold_caps_min; /* jiffies */
321 unsigned long i_hold_caps_max; /* jiffies */
322 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
323 int i_cap_exporting_mds; /* to handle cap migration between */
324 unsigned i_cap_exporting_mseq; /* mds's. */
325 unsigned i_cap_exporting_issued;
326 struct ceph_cap_reservation i_cap_migration_resv;
327 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
328 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 */
329 unsigned i_snap_caps; /* cap bits for snapped files */
331 int i_nr_by_mode[CEPH_FILE_MODE_NUM]; /* open file counts */
333 u32 i_truncate_seq; /* last truncate to smaller size */
334 u64 i_truncate_size; /* and the size we last truncated down to */
335 int i_truncate_pending; /* still need to call vmtruncate */
337 u64 i_max_size; /* max file size authorized by mds */
338 u64 i_reported_size; /* (max_)size reported to or requested of mds */
339 u64 i_wanted_max_size; /* offset we'd like to write too */
340 u64 i_requested_max_size; /* max_size we've requested */
342 /* held references to caps */
344 int i_rd_ref, i_rdcache_ref, i_wr_ref;
345 int i_wrbuffer_ref, i_wrbuffer_ref_head;
346 u32 i_shared_gen; /* increment each time we get FILE_SHARED */
347 u32 i_rdcache_gen; /* we increment this each time we get
348 FILE_CACHE. If it's non-zero, we
349 _may_ have cached pages. */
350 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
352 struct list_head i_unsafe_writes; /* uncommitted sync writes */
353 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
354 spinlock_t i_unsafe_lock;
356 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
357 int i_snap_realm_counter; /* snap realm (if caps) */
358 struct list_head i_snap_realm_item;
359 struct list_head i_snap_flush_item;
361 struct work_struct i_wb_work; /* writeback work */
362 struct work_struct i_pg_inv_work; /* page invalidation work */
364 struct work_struct i_vmtruncate_work;
366 struct inode vfs_inode; /* at end */
369 static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
371 return container_of(inode, struct ceph_inode_info, vfs_inode);
374 static inline void ceph_i_clear(struct inode *inode, unsigned mask)
376 struct ceph_inode_info *ci = ceph_inode(inode);
378 spin_lock(&inode->i_lock);
379 ci->i_ceph_flags &= ~mask;
380 spin_unlock(&inode->i_lock);
383 static inline void ceph_i_set(struct inode *inode, unsigned mask)
385 struct ceph_inode_info *ci = ceph_inode(inode);
387 spin_lock(&inode->i_lock);
388 ci->i_ceph_flags |= mask;
389 spin_unlock(&inode->i_lock);
392 static inline bool ceph_i_test(struct inode *inode, unsigned mask)
394 struct ceph_inode_info *ci = ceph_inode(inode);
398 r = (ci->i_ceph_flags & mask) == mask;
403 /* find a specific frag @f */
404 extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
408 * choose fragment for value @v. copy frag content to pfrag, if leaf
411 extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
412 struct ceph_inode_frag *pfrag,
418 struct ceph_dentry_info {
419 struct ceph_mds_session *lease_session;
420 u32 lease_gen, lease_shared_gen;
422 unsigned long lease_renew_after, lease_renew_from;
423 struct list_head lru;
424 struct dentry *dentry;
429 static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
431 return (struct ceph_dentry_info *)dentry->d_fsdata;
434 static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
436 return ((loff_t)frag << 32) | (loff_t)off;
440 * ino_t is <64 bits on many architectures, blech.
442 * don't include snap in ino hash, at least for now.
444 static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
446 ino_t ino = (ino_t)vino.ino; /* ^ (vino.snap << 20); */
447 #if BITS_PER_LONG == 32
448 ino ^= vino.ino >> (sizeof(u64)-sizeof(ino_t)) * 8;
455 static inline int ceph_set_ino_cb(struct inode *inode, void *data)
457 ceph_inode(inode)->i_vino = *(struct ceph_vino *)data;
458 inode->i_ino = ceph_vino_to_ino(*(struct ceph_vino *)data);
462 static inline struct ceph_vino ceph_vino(struct inode *inode)
464 return ceph_inode(inode)->i_vino;
467 /* for printf-style formatting */
468 #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
470 static inline u64 ceph_ino(struct inode *inode)
472 return ceph_inode(inode)->i_vino.ino;
474 static inline u64 ceph_snap(struct inode *inode)
476 return ceph_inode(inode)->i_vino.snap;
479 static inline int ceph_ino_compare(struct inode *inode, void *data)
481 struct ceph_vino *pvino = (struct ceph_vino *)data;
482 struct ceph_inode_info *ci = ceph_inode(inode);
483 return ci->i_vino.ino == pvino->ino &&
484 ci->i_vino.snap == pvino->snap;
487 static inline struct inode *ceph_find_inode(struct super_block *sb,
488 struct ceph_vino vino)
490 ino_t t = ceph_vino_to_ino(vino);
491 return ilookup5(sb, t, ceph_ino_compare, &vino);
498 static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
500 return !RB_EMPTY_ROOT(&ci->i_caps);
503 extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
504 extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
505 extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
506 struct ceph_cap *cap);
508 static inline int ceph_caps_issued(struct ceph_inode_info *ci)
511 spin_lock(&ci->vfs_inode.i_lock);
512 issued = __ceph_caps_issued(ci, NULL);
513 spin_unlock(&ci->vfs_inode.i_lock);
517 static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
521 spin_lock(&ci->vfs_inode.i_lock);
522 r = __ceph_caps_issued_mask(ci, mask, touch);
523 spin_unlock(&ci->vfs_inode.i_lock);
527 static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
529 return ci->i_dirty_caps | ci->i_flushing_caps;
531 extern void __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask);
533 extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
534 extern int __ceph_caps_used(struct ceph_inode_info *ci);
536 extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
539 * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
541 static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
543 int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
544 if (w & CEPH_CAP_FILE_BUFFER)
545 w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */
549 /* what the mds thinks we want */
550 extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
552 extern void ceph_caps_init(void);
553 extern void ceph_caps_finalize(void);
554 extern int ceph_reserve_caps(struct ceph_cap_reservation *ctx, int need);
555 extern int ceph_unreserve_caps(struct ceph_cap_reservation *ctx);
556 extern void ceph_reservation_status(struct ceph_client *client,
557 int *total, int *avail, int *used,
560 static inline struct ceph_client *ceph_inode_to_client(struct inode *inode)
562 return (struct ceph_client *)inode->i_sb->s_fs_info;
565 static inline struct ceph_client *ceph_sb_to_client(struct super_block *sb)
567 return (struct ceph_client *)sb->s_fs_info;
570 static inline int ceph_queue_writeback(struct inode *inode)
572 return queue_work(ceph_inode_to_client(inode)->wb_wq,
573 &ceph_inode(inode)->i_wb_work);
576 static inline int ceph_queue_page_invalidation(struct inode *inode)
578 return queue_work(ceph_inode_to_client(inode)->pg_inv_wq,
579 &ceph_inode(inode)->i_pg_inv_work);
584 * we keep buffered readdir results attached to file->private_data
586 struct ceph_file_info {
587 int fmode; /* initialized on open */
589 /* readdir: position within the dir */
591 struct ceph_mds_request *last_readdir;
594 /* readdir: position within a frag */
595 unsigned offset; /* offset of last chunk, adjusted for . and .. */
596 u64 next_offset; /* offset of next chunk (last_name's + 1) */
597 char *last_name; /* last entry in previous chunk */
598 struct dentry *dentry; /* next dentry (for dcache readdir) */
599 unsigned long dir_release_count;
601 /* used for -o dirstat read() on directory thing */
613 * A "snap context" is the set of existing snapshots when we
614 * write data. It is used by the OSD to guide its COW behavior.
616 * The ceph_snap_context is refcounted, and attached to each dirty
617 * page, indicating which context the dirty data belonged when it was
620 struct ceph_snap_context {
627 static inline struct ceph_snap_context *
628 ceph_get_snap_context(struct ceph_snap_context *sc)
631 printk("get_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
632 atomic_read(&sc->nref)+1);
635 atomic_inc(&sc->nref);
639 static inline void ceph_put_snap_context(struct ceph_snap_context *sc)
644 printk("put_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
645 atomic_read(&sc->nref)-1);
647 if (atomic_dec_and_test(&sc->nref)) {
648 /*printk(" deleting snap_context %p\n", sc);*/
654 * 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.
659 * All inodes within the realm that have capabilities are linked into a
662 struct ceph_snap_realm {
667 u64 parent_since; /* snapid when our current parent became so */
669 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
670 int num_prior_parent_snaps; /* had prior to parent_since */
671 u64 *snaps; /* snaps specific to this realm */
674 struct ceph_snap_realm *parent;
675 struct list_head children; /* list of child realms */
676 struct list_head child_item;
678 struct list_head empty_item; /* if i have ref==0 */
680 /* the current set of snaps for this realm */
681 struct ceph_snap_context *cached_context;
683 struct list_head inodes_with_caps;
684 spinlock_t inodes_with_caps_lock;
690 * calculate the number of pages a given length and offset map onto,
691 * if we align the data.
693 static inline int calc_pages_for(u64 off, u64 len)
695 return ((off+len+PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT) -
696 (off >> PAGE_CACHE_SHIFT);
702 struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
704 extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
705 struct ceph_snap_realm *realm);
706 extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
707 struct ceph_snap_realm *realm);
708 extern int ceph_update_snap_trace(struct ceph_mds_client *m,
709 void *p, void *e, bool deletion);
710 extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
711 struct ceph_msg *msg);
712 extern void ceph_queue_cap_snap(struct ceph_inode_info *ci,
713 struct ceph_snap_context *snapc);
714 extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
715 struct ceph_cap_snap *capsnap);
716 extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
719 * a cap_snap is "pending" if it is still awaiting an in-progress
720 * sync write (that may/may not still update size, mtime, etc.).
722 static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
724 return !list_empty(&ci->i_cap_snaps) &&
725 list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap,
731 extern struct kmem_cache *ceph_inode_cachep;
732 extern struct kmem_cache *ceph_cap_cachep;
733 extern struct kmem_cache *ceph_dentry_cachep;
734 extern struct kmem_cache *ceph_file_cachep;
736 extern const char *ceph_msg_type_name(int type);
737 extern int ceph_check_fsid(struct ceph_client *client, struct ceph_fsid *fsid);
739 #define FSID_FORMAT "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-" \
740 "%02x%02x%02x%02x%02x%02x"
741 #define PR_FSID(f) (f)->fsid[0], (f)->fsid[1], (f)->fsid[2], (f)->fsid[3], \
742 (f)->fsid[4], (f)->fsid[5], (f)->fsid[6], (f)->fsid[7], \
743 (f)->fsid[8], (f)->fsid[9], (f)->fsid[10], (f)->fsid[11], \
744 (f)->fsid[12], (f)->fsid[13], (f)->fsid[14], (f)->fsid[15]
747 extern const struct inode_operations ceph_file_iops;
749 extern struct inode *ceph_alloc_inode(struct super_block *sb);
750 extern void ceph_destroy_inode(struct inode *inode);
752 extern struct inode *ceph_get_inode(struct super_block *sb,
753 struct ceph_vino vino);
754 extern struct inode *ceph_get_snapdir(struct inode *parent);
755 extern int ceph_fill_file_size(struct inode *inode, int issued,
756 u32 truncate_seq, u64 truncate_size, u64 size);
757 extern void ceph_fill_file_time(struct inode *inode, int issued,
758 u64 time_warp_seq, struct timespec *ctime,
759 struct timespec *mtime, struct timespec *atime);
760 extern int ceph_fill_trace(struct super_block *sb,
761 struct ceph_mds_request *req,
762 struct ceph_mds_session *session);
763 extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
764 struct ceph_mds_session *session);
766 extern int ceph_inode_holds_cap(struct inode *inode, int mask);
768 extern int ceph_inode_set_size(struct inode *inode, loff_t size);
769 extern void ceph_inode_writeback(struct work_struct *work);
770 extern void ceph_vmtruncate_work(struct work_struct *work);
771 extern void __ceph_do_pending_vmtruncate(struct inode *inode);
772 extern void __ceph_queue_vmtruncate(struct inode *inode);
774 extern int ceph_do_getattr(struct inode *inode, int mask);
775 extern int ceph_permission(struct inode *inode, int mask);
776 extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
777 extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
781 extern int ceph_setxattr(struct dentry *, const char *, const void *,
783 extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t);
784 extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
785 extern int ceph_removexattr(struct dentry *, const char *);
786 extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
787 extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
790 extern const char *ceph_cap_string(int c);
791 extern void ceph_handle_caps(struct ceph_mds_session *session,
792 struct ceph_msg *msg);
793 extern int ceph_add_cap(struct inode *inode,
794 struct ceph_mds_session *session, u64 cap_id,
795 int fmode, unsigned issued, unsigned wanted,
796 unsigned cap, unsigned seq, u64 realmino, int flags,
797 struct ceph_cap_reservation *caps_reservation);
798 extern void __ceph_remove_cap(struct ceph_cap *cap,
799 struct ceph_cap_reservation *ctx);
800 static inline void ceph_remove_cap(struct ceph_cap *cap)
802 struct inode *inode = &cap->ci->vfs_inode;
803 spin_lock(&inode->i_lock);
804 __ceph_remove_cap(cap, NULL);
805 spin_unlock(&inode->i_lock);
808 extern void ceph_queue_caps_release(struct inode *inode);
809 extern int ceph_write_inode(struct inode *inode, int unused);
810 extern int ceph_fsync(struct file *file, struct dentry *dentry, int datasync);
811 extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
812 struct ceph_mds_session *session);
813 extern int ceph_get_cap_mds(struct inode *inode);
814 extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
815 extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
816 extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
817 struct ceph_snap_context *snapc);
818 extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
819 struct ceph_mds_session **psession);
820 extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
821 struct ceph_mds_session *session);
822 extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
823 extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
825 extern int ceph_encode_inode_release(void **p, struct inode *inode,
826 int mds, int drop, int unless, int force);
827 extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
828 int mds, int drop, int unless);
830 extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
831 int *got, loff_t endoff);
833 /* for counting open files by mode */
834 static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
836 ci->i_nr_by_mode[mode]++;
838 extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
841 extern const struct address_space_operations ceph_aops;
842 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
845 extern const struct file_operations ceph_file_fops;
846 extern const struct address_space_operations ceph_aops;
847 extern int ceph_open(struct inode *inode, struct file *file);
848 extern struct dentry *ceph_lookup_open(struct inode *dir, struct dentry *dentry,
849 struct nameidata *nd, int mode,
851 extern int ceph_release(struct inode *inode, struct file *filp);
852 extern void ceph_release_page_vector(struct page **pages, int num_pages);
855 extern const struct file_operations ceph_dir_fops;
856 extern const struct inode_operations ceph_dir_iops;
857 extern struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
858 ceph_snapdir_dentry_ops;
860 extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
861 extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
862 struct dentry *dentry, int err);
864 extern void ceph_dentry_lru_add(struct dentry *dn);
865 extern void ceph_dentry_lru_touch(struct dentry *dn);
866 extern void ceph_dentry_lru_del(struct dentry *dn);
869 * our d_ops vary depending on whether the inode is live,
870 * snapshotted (read-only), or a virtual ".snap" directory.
872 int ceph_init_dentry(struct dentry *dentry);
876 extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
879 extern const struct export_operations ceph_export_ops;
882 extern int ceph_debugfs_init(void);
883 extern void ceph_debugfs_cleanup(void);
884 extern int ceph_debugfs_client_init(struct ceph_client *client);
885 extern void ceph_debugfs_client_cleanup(struct ceph_client *client);
887 static inline struct inode *get_dentry_parent_inode(struct dentry *dentry)
889 if (dentry && dentry->d_parent)
890 return dentry->d_parent->d_inode;
895 #endif /* _FS_CEPH_SUPER_H */