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 */
62 int congestion_kb; /* max readdir size */
64 char *snapdir_name; /* default ".snap" */
67 int cap_release_safety;
73 #define CEPH_MOUNT_TIMEOUT_DEFAULT 60
74 #define CEPH_MOUNT_RSIZE_DEFAULT (512*1024) /* readahead */
76 #define CEPH_MSG_MAX_FRONT_LEN (16*1024*1024)
77 #define CEPH_MSG_MAX_DATA_LEN (16*1024*1024)
79 #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
80 #define CEPH_AUTH_NAME_DEFAULT "guest"
83 * Delay telling the MDS we no longer want caps, in case we reopen
84 * the file. Delay a minimum amount of time, even if we send a cap
85 * message for some other reason. Otherwise, take the oppotunity to
86 * update the mds to avoid sending another message later.
88 #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */
89 #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */
96 CEPH_MOUNT_UNMOUNTING,
104 static inline unsigned long time_sub(unsigned long a, unsigned long b)
106 BUG_ON(time_after(b, a));
107 return (long)a - (long)b;
111 * per-filesystem client state
113 * possibly shared by multiple mount points, if they are
114 * mounting the same ceph filesystem/cluster.
117 struct ceph_fsid fsid;
120 struct mutex mount_mutex; /* serialize mount attempts */
121 struct ceph_mount_args *mount_args;
123 struct super_block *sb;
125 unsigned long mount_state;
126 wait_queue_head_t auth_wq;
130 struct ceph_messenger *msgr; /* messenger instance */
131 struct ceph_mon_client monc;
132 struct ceph_mds_client mdsc;
133 struct ceph_osd_client osdc;
136 mempool_t *wb_pagevec_pool;
137 struct workqueue_struct *wb_wq;
138 struct workqueue_struct *pg_inv_wq;
139 struct workqueue_struct *trunc_wq;
140 atomic_long_t writeback_count;
142 struct backing_dev_info backing_dev_info;
144 #ifdef CONFIG_DEBUG_FS
145 struct dentry *debugfs_monmap;
146 struct dentry *debugfs_mdsmap, *debugfs_osdmap;
147 struct dentry *debugfs_dir, *debugfs_dentry_lru, *debugfs_caps;
148 struct dentry *debugfs_congestion_kb;
149 struct dentry *debugfs_bdi;
153 static inline struct ceph_client *ceph_client(struct super_block *sb)
155 return sb->s_fs_info;
160 * File i/o capability. This tracks shared state with the metadata
161 * server that allows us to cache or writeback attributes or to read
162 * and write data. For any given inode, we should have one or more
163 * capabilities, one issued by each metadata server, and our
164 * cumulative access is the OR of all issued capabilities.
166 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
167 * session capability lists.
170 struct ceph_inode_info *ci;
171 struct rb_node ci_node; /* per-ci cap tree */
172 struct ceph_mds_session *session;
173 struct list_head session_caps; /* per-session caplist */
175 u64 cap_id; /* unique cap id (mds provided) */
176 int issued; /* latest, from the mds */
177 int implemented; /* implemented superset of issued (for revocation) */
179 u32 seq, issue_seq, mseq;
180 u32 cap_gen; /* active/stale cycle */
181 unsigned long last_used;
182 struct list_head caps_item;
185 #define CHECK_CAPS_NODELAY 1 /* do not delay any further */
186 #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */
187 #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */
190 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
191 * we first complete any in-process sync writes and writeback any dirty
192 * data before flushing the snapped state (tracked here) back to the MDS.
194 struct ceph_cap_snap {
196 struct ceph_inode_info *ci;
197 struct list_head ci_item, flushing_item;
199 u64 follows, flush_tid;
201 struct ceph_snap_context *context;
212 struct timespec mtime, atime, ctime;
214 int writing; /* a sync write is still in progress */
215 int dirty_pages; /* dirty pages awaiting writeback */
218 static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
220 if (atomic_dec_and_test(&capsnap->nref))
225 * The frag tree describes how a directory is fragmented, potentially across
226 * multiple metadata servers. It is also used to indicate points where
227 * metadata authority is delegated, and whether/where metadata is replicated.
229 * A _leaf_ frag will be present in the i_fragtree IFF there is
230 * delegation info. That is, if mds >= 0 || ndist > 0.
232 #define CEPH_MAX_DIRFRAG_REP 4
234 struct ceph_inode_frag {
239 int split_by; /* i.e. 2^(split_by) children */
241 /* delegation and replication info */
242 int mds; /* -1 if same authority as parent */
243 int ndist; /* >0 if replicated */
244 int dist[CEPH_MAX_DIRFRAG_REP];
248 * We cache inode xattrs as an encoded blob until they are first used,
249 * at which point we parse them into an rbtree.
251 struct ceph_inode_xattr {
260 int should_free_name;
264 struct ceph_inode_xattrs_info {
266 * (still encoded) xattr blob. we avoid the overhead of parsing
267 * this until someone actually calls getxattr, etc.
269 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
270 * NULL means we don't know.
272 struct ceph_buffer *blob, *prealloc_blob;
274 struct rb_root index;
279 u64 version, index_version;
285 #define CEPH_I_COMPLETE 1 /* we have complete directory cached */
286 #define CEPH_I_NODELAY 4 /* do not delay cap release */
287 #define CEPH_I_FLUSH 8 /* do not delay flush of dirty metadata */
289 struct ceph_inode_info {
290 struct ceph_vino i_vino; /* ceph ino + snap */
295 unsigned i_ceph_flags;
296 unsigned long i_release_count;
298 struct ceph_file_layout i_layout;
302 struct timespec i_rctime;
303 u64 i_rbytes, i_rfiles, i_rsubdirs;
304 u64 i_files, i_subdirs;
305 u64 i_max_offset; /* largest readdir offset, set with I_COMPLETE */
307 struct rb_root i_fragtree;
308 struct mutex i_fragtree_mutex;
310 struct ceph_inode_xattrs_info i_xattrs;
312 /* capabilities. protected _both_ by i_lock and cap->session's
314 struct rb_root i_caps; /* cap list */
315 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
316 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
317 struct list_head i_dirty_item, i_flushing_item;
319 /* we need to track cap writeback on a per-cap-bit basis, to allow
320 * overlapping, pipelined cap flushes to the mds. we can probably
321 * reduce the tid to 8 bits if we're concerned about inode size. */
322 u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS];
323 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
324 unsigned long i_hold_caps_min; /* jiffies */
325 unsigned long i_hold_caps_max; /* jiffies */
326 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
327 int i_cap_exporting_mds; /* to handle cap migration between */
328 unsigned i_cap_exporting_mseq; /* mds's. */
329 unsigned i_cap_exporting_issued;
330 struct ceph_cap_reservation i_cap_migration_resv;
331 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
332 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 */
333 unsigned i_snap_caps; /* cap bits for snapped files */
335 int i_nr_by_mode[CEPH_FILE_MODE_NUM]; /* open file counts */
337 u32 i_truncate_seq; /* last truncate to smaller size */
338 u64 i_truncate_size; /* and the size we last truncated down to */
339 int i_truncate_pending; /* still need to call vmtruncate */
341 u64 i_max_size; /* max file size authorized by mds */
342 u64 i_reported_size; /* (max_)size reported to or requested of mds */
343 u64 i_wanted_max_size; /* offset we'd like to write too */
344 u64 i_requested_max_size; /* max_size we've requested */
346 /* held references to caps */
348 int i_rd_ref, i_rdcache_ref, i_wr_ref;
349 int i_wrbuffer_ref, i_wrbuffer_ref_head;
350 u32 i_shared_gen; /* increment each time we get FILE_SHARED */
351 u32 i_rdcache_gen; /* we increment this each time we get
352 FILE_CACHE. If it's non-zero, we
353 _may_ have cached pages. */
354 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
356 struct list_head i_unsafe_writes; /* uncommitted sync writes */
357 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
358 spinlock_t i_unsafe_lock;
360 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
361 int i_snap_realm_counter; /* snap realm (if caps) */
362 struct list_head i_snap_realm_item;
363 struct list_head i_snap_flush_item;
365 struct work_struct i_wb_work; /* writeback work */
366 struct work_struct i_pg_inv_work; /* page invalidation work */
368 struct work_struct i_vmtruncate_work;
370 struct inode vfs_inode; /* at end */
373 static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
375 return container_of(inode, struct ceph_inode_info, vfs_inode);
378 static inline void ceph_i_clear(struct inode *inode, unsigned mask)
380 struct ceph_inode_info *ci = ceph_inode(inode);
382 spin_lock(&inode->i_lock);
383 ci->i_ceph_flags &= ~mask;
384 spin_unlock(&inode->i_lock);
387 static inline void ceph_i_set(struct inode *inode, unsigned mask)
389 struct ceph_inode_info *ci = ceph_inode(inode);
391 spin_lock(&inode->i_lock);
392 ci->i_ceph_flags |= mask;
393 spin_unlock(&inode->i_lock);
396 static inline bool ceph_i_test(struct inode *inode, unsigned mask)
398 struct ceph_inode_info *ci = ceph_inode(inode);
402 r = (ci->i_ceph_flags & mask) == mask;
407 /* find a specific frag @f */
408 extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
412 * choose fragment for value @v. copy frag content to pfrag, if leaf
415 extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
416 struct ceph_inode_frag *pfrag,
422 struct ceph_dentry_info {
423 struct ceph_mds_session *lease_session;
424 u32 lease_gen, lease_shared_gen;
426 unsigned long lease_renew_after, lease_renew_from;
427 struct list_head lru;
428 struct dentry *dentry;
433 static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
435 return (struct ceph_dentry_info *)dentry->d_fsdata;
438 static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
440 return ((loff_t)frag << 32) | (loff_t)off;
444 * ino_t is <64 bits on many architectures, blech.
446 * don't include snap in ino hash, at least for now.
448 static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
450 ino_t ino = (ino_t)vino.ino; /* ^ (vino.snap << 20); */
451 #if BITS_PER_LONG == 32
452 ino ^= vino.ino >> (sizeof(u64)-sizeof(ino_t)) * 8;
459 static inline int ceph_set_ino_cb(struct inode *inode, void *data)
461 ceph_inode(inode)->i_vino = *(struct ceph_vino *)data;
462 inode->i_ino = ceph_vino_to_ino(*(struct ceph_vino *)data);
466 static inline struct ceph_vino ceph_vino(struct inode *inode)
468 return ceph_inode(inode)->i_vino;
471 /* for printf-style formatting */
472 #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
474 static inline u64 ceph_ino(struct inode *inode)
476 return ceph_inode(inode)->i_vino.ino;
478 static inline u64 ceph_snap(struct inode *inode)
480 return ceph_inode(inode)->i_vino.snap;
483 static inline int ceph_ino_compare(struct inode *inode, void *data)
485 struct ceph_vino *pvino = (struct ceph_vino *)data;
486 struct ceph_inode_info *ci = ceph_inode(inode);
487 return ci->i_vino.ino == pvino->ino &&
488 ci->i_vino.snap == pvino->snap;
491 static inline struct inode *ceph_find_inode(struct super_block *sb,
492 struct ceph_vino vino)
494 ino_t t = ceph_vino_to_ino(vino);
495 return ilookup5(sb, t, ceph_ino_compare, &vino);
502 static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
504 return !RB_EMPTY_ROOT(&ci->i_caps);
507 extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
508 extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
509 extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
510 struct ceph_cap *cap);
512 static inline int ceph_caps_issued(struct ceph_inode_info *ci)
515 spin_lock(&ci->vfs_inode.i_lock);
516 issued = __ceph_caps_issued(ci, NULL);
517 spin_unlock(&ci->vfs_inode.i_lock);
521 static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
525 spin_lock(&ci->vfs_inode.i_lock);
526 r = __ceph_caps_issued_mask(ci, mask, touch);
527 spin_unlock(&ci->vfs_inode.i_lock);
531 static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
533 return ci->i_dirty_caps | ci->i_flushing_caps;
535 extern void __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask);
537 extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
538 extern int __ceph_caps_used(struct ceph_inode_info *ci);
540 extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
543 * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
545 static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
547 int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
548 if (w & CEPH_CAP_FILE_BUFFER)
549 w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */
553 /* what the mds thinks we want */
554 extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
556 extern void ceph_caps_init(void);
557 extern void ceph_caps_finalize(void);
558 extern int ceph_reserve_caps(struct ceph_cap_reservation *ctx, int need);
559 extern int ceph_unreserve_caps(struct ceph_cap_reservation *ctx);
560 extern void ceph_reservation_status(struct ceph_client *client,
561 int *total, int *avail, int *used,
564 static inline struct ceph_client *ceph_inode_to_client(struct inode *inode)
566 return (struct ceph_client *)inode->i_sb->s_fs_info;
569 static inline struct ceph_client *ceph_sb_to_client(struct super_block *sb)
571 return (struct ceph_client *)sb->s_fs_info;
574 static inline int ceph_queue_writeback(struct inode *inode)
576 return queue_work(ceph_inode_to_client(inode)->wb_wq,
577 &ceph_inode(inode)->i_wb_work);
580 static inline int ceph_queue_page_invalidation(struct inode *inode)
582 return queue_work(ceph_inode_to_client(inode)->pg_inv_wq,
583 &ceph_inode(inode)->i_pg_inv_work);
588 * we keep buffered readdir results attached to file->private_data
590 struct ceph_file_info {
591 int fmode; /* initialized on open */
593 /* readdir: position within the dir */
595 struct ceph_mds_request *last_readdir;
598 /* readdir: position within a frag */
599 unsigned offset; /* offset of last chunk, adjusted for . and .. */
600 u64 next_offset; /* offset of next chunk (last_name's + 1) */
601 char *last_name; /* last entry in previous chunk */
602 struct dentry *dentry; /* next dentry (for dcache readdir) */
603 unsigned long dir_release_count;
605 /* used for -o dirstat read() on directory thing */
617 * A "snap context" is the set of existing snapshots when we
618 * write data. It is used by the OSD to guide its COW behavior.
620 * The ceph_snap_context is refcounted, and attached to each dirty
621 * page, indicating which context the dirty data belonged when it was
624 struct ceph_snap_context {
631 static inline struct ceph_snap_context *
632 ceph_get_snap_context(struct ceph_snap_context *sc)
635 printk("get_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
636 atomic_read(&sc->nref)+1);
639 atomic_inc(&sc->nref);
643 static inline void ceph_put_snap_context(struct ceph_snap_context *sc)
648 printk("put_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
649 atomic_read(&sc->nref)-1);
651 if (atomic_dec_and_test(&sc->nref)) {
652 /*printk(" deleting snap_context %p\n", sc);*/
658 * A "snap realm" describes a subset of the file hierarchy sharing
659 * the same set of snapshots that apply to it. The realms themselves
660 * are organized into a hierarchy, such that children inherit (some of)
661 * the snapshots of their parents.
663 * All inodes within the realm that have capabilities are linked into a
666 struct ceph_snap_realm {
671 u64 parent_since; /* snapid when our current parent became so */
673 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
674 int num_prior_parent_snaps; /* had prior to parent_since */
675 u64 *snaps; /* snaps specific to this realm */
678 struct ceph_snap_realm *parent;
679 struct list_head children; /* list of child realms */
680 struct list_head child_item;
682 struct list_head empty_item; /* if i have ref==0 */
684 /* the current set of snaps for this realm */
685 struct ceph_snap_context *cached_context;
687 struct list_head inodes_with_caps;
688 spinlock_t inodes_with_caps_lock;
694 * calculate the number of pages a given length and offset map onto,
695 * if we align the data.
697 static inline int calc_pages_for(u64 off, u64 len)
699 return ((off+len+PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT) -
700 (off >> PAGE_CACHE_SHIFT);
706 struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
708 extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
709 struct ceph_snap_realm *realm);
710 extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
711 struct ceph_snap_realm *realm);
712 extern int ceph_update_snap_trace(struct ceph_mds_client *m,
713 void *p, void *e, bool deletion);
714 extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
715 struct ceph_msg *msg);
716 extern void ceph_queue_cap_snap(struct ceph_inode_info *ci,
717 struct ceph_snap_context *snapc);
718 extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
719 struct ceph_cap_snap *capsnap);
720 extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
723 * a cap_snap is "pending" if it is still awaiting an in-progress
724 * sync write (that may/may not still update size, mtime, etc.).
726 static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
728 return !list_empty(&ci->i_cap_snaps) &&
729 list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap,
735 extern struct kmem_cache *ceph_inode_cachep;
736 extern struct kmem_cache *ceph_cap_cachep;
737 extern struct kmem_cache *ceph_dentry_cachep;
738 extern struct kmem_cache *ceph_file_cachep;
740 extern const char *ceph_msg_type_name(int type);
741 extern int ceph_check_fsid(struct ceph_client *client, struct ceph_fsid *fsid);
743 #define FSID_FORMAT "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-" \
744 "%02x%02x%02x%02x%02x%02x"
745 #define PR_FSID(f) (f)->fsid[0], (f)->fsid[1], (f)->fsid[2], (f)->fsid[3], \
746 (f)->fsid[4], (f)->fsid[5], (f)->fsid[6], (f)->fsid[7], \
747 (f)->fsid[8], (f)->fsid[9], (f)->fsid[10], (f)->fsid[11], \
748 (f)->fsid[12], (f)->fsid[13], (f)->fsid[14], (f)->fsid[15]
751 extern const struct inode_operations ceph_file_iops;
753 extern struct inode *ceph_alloc_inode(struct super_block *sb);
754 extern void ceph_destroy_inode(struct inode *inode);
756 extern struct inode *ceph_get_inode(struct super_block *sb,
757 struct ceph_vino vino);
758 extern struct inode *ceph_get_snapdir(struct inode *parent);
759 extern int ceph_fill_file_size(struct inode *inode, int issued,
760 u32 truncate_seq, u64 truncate_size, u64 size);
761 extern void ceph_fill_file_time(struct inode *inode, int issued,
762 u64 time_warp_seq, struct timespec *ctime,
763 struct timespec *mtime, struct timespec *atime);
764 extern int ceph_fill_trace(struct super_block *sb,
765 struct ceph_mds_request *req,
766 struct ceph_mds_session *session);
767 extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
768 struct ceph_mds_session *session);
770 extern int ceph_inode_holds_cap(struct inode *inode, int mask);
772 extern int ceph_inode_set_size(struct inode *inode, loff_t size);
773 extern void ceph_inode_writeback(struct work_struct *work);
774 extern void ceph_vmtruncate_work(struct work_struct *work);
775 extern void __ceph_do_pending_vmtruncate(struct inode *inode);
776 extern void __ceph_queue_vmtruncate(struct inode *inode);
778 extern int ceph_do_getattr(struct inode *inode, int mask);
779 extern int ceph_permission(struct inode *inode, int mask);
780 extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
781 extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
785 extern int ceph_setxattr(struct dentry *, const char *, const void *,
787 extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t);
788 extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
789 extern int ceph_removexattr(struct dentry *, const char *);
790 extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
791 extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
794 extern const char *ceph_cap_string(int c);
795 extern void ceph_handle_caps(struct ceph_mds_session *session,
796 struct ceph_msg *msg);
797 extern int ceph_add_cap(struct inode *inode,
798 struct ceph_mds_session *session, u64 cap_id,
799 int fmode, unsigned issued, unsigned wanted,
800 unsigned cap, unsigned seq, u64 realmino, int flags,
801 struct ceph_cap_reservation *caps_reservation);
802 extern void __ceph_remove_cap(struct ceph_cap *cap,
803 struct ceph_cap_reservation *ctx);
804 static inline void ceph_remove_cap(struct ceph_cap *cap)
806 struct inode *inode = &cap->ci->vfs_inode;
807 spin_lock(&inode->i_lock);
808 __ceph_remove_cap(cap, NULL);
809 spin_unlock(&inode->i_lock);
812 extern void ceph_queue_caps_release(struct inode *inode);
813 extern int ceph_write_inode(struct inode *inode, int unused);
814 extern int ceph_fsync(struct file *file, struct dentry *dentry, int datasync);
815 extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
816 struct ceph_mds_session *session);
817 extern int ceph_get_cap_mds(struct inode *inode);
818 extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
819 extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
820 extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
821 struct ceph_snap_context *snapc);
822 extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
823 struct ceph_mds_session **psession);
824 extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
825 struct ceph_mds_session *session);
826 extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
827 extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
829 extern int ceph_encode_inode_release(void **p, struct inode *inode,
830 int mds, int drop, int unless, int force);
831 extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
832 int mds, int drop, int unless);
834 extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
835 int *got, loff_t endoff);
837 /* for counting open files by mode */
838 static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
840 ci->i_nr_by_mode[mode]++;
842 extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
845 extern const struct address_space_operations ceph_aops;
846 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
849 extern const struct file_operations ceph_file_fops;
850 extern const struct address_space_operations ceph_aops;
851 extern int ceph_open(struct inode *inode, struct file *file);
852 extern struct dentry *ceph_lookup_open(struct inode *dir, struct dentry *dentry,
853 struct nameidata *nd, int mode,
855 extern int ceph_release(struct inode *inode, struct file *filp);
856 extern void ceph_release_page_vector(struct page **pages, int num_pages);
859 extern const struct file_operations ceph_dir_fops;
860 extern const struct inode_operations ceph_dir_iops;
861 extern struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
862 ceph_snapdir_dentry_ops;
864 extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
865 extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
866 struct dentry *dentry, int err);
868 extern void ceph_dentry_lru_add(struct dentry *dn);
869 extern void ceph_dentry_lru_touch(struct dentry *dn);
870 extern void ceph_dentry_lru_del(struct dentry *dn);
873 * our d_ops vary depending on whether the inode is live,
874 * snapshotted (read-only), or a virtual ".snap" directory.
876 int ceph_init_dentry(struct dentry *dentry);
880 extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
883 extern const struct export_operations ceph_export_ops;
886 extern int ceph_debugfs_init(void);
887 extern void ceph_debugfs_cleanup(void);
888 extern int ceph_debugfs_client_init(struct ceph_client *client);
889 extern void ceph_debugfs_client_cleanup(struct ceph_client *client);
891 static inline struct inode *get_dentry_parent_inode(struct dentry *dentry)
893 if (dentry && dentry->d_parent)
894 return dentry->d_parent->d_inode;
899 #endif /* _FS_CEPH_SUPER_H */