1 #ifndef _FS_CEPH_SUPER_H
2 #define _FS_CEPH_SUPER_H
4 #include <linux/ceph/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>
14 #include <linux/writeback.h>
15 #include <linux/slab.h>
16 #include <linux/posix_acl.h>
18 #include <linux/ceph/libceph.h>
20 #ifdef CONFIG_CEPH_FSCACHE
21 #include <linux/fscache.h>
24 /* f_type in struct statfs */
25 #define CEPH_SUPER_MAGIC 0x00c36400
27 /* large granularity for statfs utilization stats to facilitate
28 * large volume sizes on 32-bit machines. */
29 #define CEPH_BLOCK_SHIFT 22 /* 4 MB */
30 #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT)
32 #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 */
35 #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */
36 #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */
37 #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */
38 #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */
40 #define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE
42 #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))
47 #define CEPH_RSIZE_DEFAULT 0 /* max read size */
48 #define CEPH_RASIZE_DEFAULT (8192*1024) /* readahead */
49 #define CEPH_MAX_READDIR_DEFAULT 1024
50 #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024)
51 #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
53 struct ceph_mount_options {
57 int wsize; /* max write size */
58 int rsize; /* max read size */
59 int rasize; /* max readahead */
60 int congestion_kb; /* max writeback in flight */
61 int caps_wanted_delay_min, caps_wanted_delay_max;
62 int cap_release_safety;
63 int max_readdir; /* max readdir result (entires) */
64 int max_readdir_bytes; /* max readdir result (bytes) */
68 * everything above this point can be memcmp'd; everything below
69 * is handled in compare_mount_options()
72 char *snapdir_name; /* default ".snap" */
73 char *server_path; /* default "/" */
76 struct ceph_fs_client {
77 struct super_block *sb;
79 struct ceph_mount_options *mount_options;
80 struct ceph_client *client;
82 unsigned long mount_state;
83 int min_caps; /* min caps i added */
85 struct ceph_mds_client *mdsc;
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;
92 atomic_long_t writeback_count;
94 struct backing_dev_info backing_dev_info;
96 #ifdef CONFIG_DEBUG_FS
97 struct dentry *debugfs_dentry_lru, *debugfs_caps;
98 struct dentry *debugfs_congestion_kb;
99 struct dentry *debugfs_bdi;
100 struct dentry *debugfs_mdsc, *debugfs_mdsmap;
101 struct dentry *debugfs_mds_sessions;
104 #ifdef CONFIG_CEPH_FSCACHE
105 struct fscache_cookie *fscache;
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.
117 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
118 * session capability lists.
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 */
125 u64 cap_id; /* unique cap id (mds provided) */
129 int issued; /* latest, from the mds */
130 int implemented; /* implemented superset of
131 issued (for revocation) */
134 /* caps to release */
140 u32 seq, issue_seq, mseq;
141 u32 cap_gen; /* active/stale cycle */
142 unsigned long last_used;
143 struct list_head caps_item;
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 */
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.
155 struct ceph_cap_snap {
157 struct ceph_inode_info *ci;
158 struct list_head ci_item, flushing_item;
160 u64 follows, flush_tid;
162 struct ceph_snap_context *context;
168 struct ceph_buffer *xattr_blob;
172 struct timespec mtime, atime, ctime;
174 int writing; /* a sync write is still in progress */
175 int dirty_pages; /* dirty pages awaiting writeback */
180 static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
182 if (atomic_dec_and_test(&capsnap->nref)) {
183 if (capsnap->xattr_blob)
184 ceph_buffer_put(capsnap->xattr_blob);
189 struct ceph_cap_flush {
192 struct rb_node g_node; // global
194 struct rb_node i_node; // inode
195 struct list_head list;
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.
204 * A _leaf_ frag will be present in the i_fragtree IFF there is
205 * delegation info. That is, if mds >= 0 || ndist > 0.
207 #define CEPH_MAX_DIRFRAG_REP 4
209 struct ceph_inode_frag {
214 int split_by; /* i.e. 2^(split_by) children */
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];
223 * We cache inode xattrs as an encoded blob until they are first used,
224 * at which point we parse them into an rbtree.
226 struct ceph_inode_xattr {
235 int should_free_name;
242 struct ceph_dentry_info {
243 struct ceph_mds_session *lease_session;
244 u32 lease_gen, lease_shared_gen;
246 unsigned long lease_renew_after, lease_renew_from;
247 struct list_head lru;
248 struct dentry *dentry;
253 struct ceph_inode_xattrs_info {
255 * (still encoded) xattr blob. we avoid the overhead of parsing
256 * this until someone actually calls getxattr, etc.
258 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
259 * NULL means we don't know.
261 struct ceph_buffer *blob, *prealloc_blob;
263 struct rb_root index;
268 u64 version, index_version;
274 struct ceph_inode_info {
275 struct ceph_vino i_vino; /* ceph ino + snap */
277 spinlock_t i_ceph_lock;
280 u64 i_inline_version;
283 unsigned i_ceph_flags;
284 atomic64_t i_release_count;
285 atomic64_t i_ordered_count;
286 atomic64_t i_complete_seq[2];
288 struct ceph_dir_layout i_dir_layout;
289 struct ceph_file_layout i_layout;
290 size_t i_pool_ns_len;
294 struct timespec i_rctime;
295 u64 i_rbytes, i_rfiles, i_rsubdirs;
296 u64 i_files, i_subdirs;
298 struct rb_root i_fragtree;
299 int i_fragtree_nsplits;
300 struct mutex i_fragtree_mutex;
302 struct ceph_inode_xattrs_info i_xattrs;
304 /* capabilities. protected _both_ by i_ceph_lock and cap->session's
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;
310 /* 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. */
313 struct ceph_cap_flush *i_prealloc_cap_flush;
314 struct rb_root i_cap_flush_tree;
315 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 */
319 struct ceph_cap_reservation i_cap_migration_resv;
320 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
321 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or
322 dirty|flushing caps */
323 unsigned i_snap_caps; /* cap bits for snapped files */
325 int i_nr_by_mode[CEPH_FILE_MODE_NUM]; /* open file counts */
327 struct mutex i_truncate_mutex;
328 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 */
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 */
337 /* held references to caps */
339 int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref;
340 int i_wrbuffer_ref, i_wrbuffer_ref_head;
341 u32 i_shared_gen; /* increment each time we get FILE_SHARED */
342 u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */
343 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
345 struct list_head i_unsafe_writes; /* uncommitted sync writes */
346 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
347 struct list_head i_unsafe_iops; /* uncommitted mds inode ops */
348 spinlock_t i_unsafe_lock;
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;
355 struct work_struct i_wb_work; /* writeback work */
356 struct work_struct i_pg_inv_work; /* page invalidation work */
358 struct work_struct i_vmtruncate_work;
360 #ifdef CONFIG_CEPH_FSCACHE
361 struct fscache_cookie *fscache;
364 struct inode vfs_inode; /* at end */
367 static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
369 return container_of(inode, struct ceph_inode_info, vfs_inode);
372 static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode)
374 return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
377 static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb)
379 return (struct ceph_fs_client *)sb->s_fs_info;
382 static inline struct ceph_vino ceph_vino(struct inode *inode)
384 return ceph_inode(inode)->i_vino;
388 * ino_t is <64 bits on many architectures, blech.
390 * i_ino (kernel inode) st_ino (userspace)
395 static inline u32 ceph_ino_to_ino32(__u64 vino)
397 u32 ino = vino & 0xffffffff;
407 static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
409 #if BITS_PER_LONG == 32
410 return ceph_ino_to_ino32(vino.ino);
412 return (ino_t)vino.ino;
417 * user-visible ino (stat, filldir)
419 #if BITS_PER_LONG == 32
420 static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
425 static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
427 if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))
428 ino = ceph_ino_to_ino32(ino);
434 /* for printf-style formatting */
435 #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
437 static inline u64 ceph_ino(struct inode *inode)
439 return ceph_inode(inode)->i_vino.ino;
441 static inline u64 ceph_snap(struct inode *inode)
443 return ceph_inode(inode)->i_vino.snap;
446 static inline int ceph_ino_compare(struct inode *inode, void *data)
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;
454 static inline struct inode *ceph_find_inode(struct super_block *sb,
455 struct ceph_vino vino)
457 ino_t t = ceph_vino_to_ino(vino);
458 return ilookup5(sb, t, ceph_ino_compare, &vino);
465 #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 */
472 #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */
473 #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */
475 static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
476 long long release_count,
477 long long ordered_count)
479 smp_mb__before_atomic();
480 atomic64_set(&ci->i_complete_seq[0], release_count);
481 atomic64_set(&ci->i_complete_seq[1], ordered_count);
484 static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
486 atomic64_inc(&ci->i_release_count);
489 static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
491 atomic64_inc(&ci->i_ordered_count);
494 static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
496 return atomic64_read(&ci->i_complete_seq[0]) ==
497 atomic64_read(&ci->i_release_count);
500 static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
502 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);
508 static inline void ceph_dir_clear_complete(struct inode *inode)
510 __ceph_dir_clear_complete(ceph_inode(inode));
513 static inline void ceph_dir_clear_ordered(struct inode *inode)
515 __ceph_dir_clear_ordered(ceph_inode(inode));
518 static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
520 bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
525 /* find a specific frag @f */
526 extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
530 * choose fragment for value @v. copy frag content to pfrag, if leaf
533 extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
534 struct ceph_inode_frag *pfrag,
537 static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
539 return (struct ceph_dentry_info *)dentry->d_fsdata;
545 static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
547 return !RB_EMPTY_ROOT(&ci->i_caps);
550 extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
551 extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
552 extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
553 struct ceph_cap *cap);
555 static inline int ceph_caps_issued(struct ceph_inode_info *ci)
558 spin_lock(&ci->i_ceph_lock);
559 issued = __ceph_caps_issued(ci, NULL);
560 spin_unlock(&ci->i_ceph_lock);
564 static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
568 spin_lock(&ci->i_ceph_lock);
569 r = __ceph_caps_issued_mask(ci, mask, touch);
570 spin_unlock(&ci->i_ceph_lock);
574 static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
576 return ci->i_dirty_caps | ci->i_flushing_caps;
578 extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
579 extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
580 extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
581 struct ceph_cap_flush **pcf);
583 extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
584 struct ceph_cap *ocap, int mask);
585 extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
586 extern int __ceph_caps_used(struct ceph_inode_info *ci);
588 extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
591 * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
593 static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
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 */
601 /* what the mds thinks we want */
602 extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
604 extern void ceph_caps_init(struct ceph_mds_client *mdsc);
605 extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
606 extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
607 extern void ceph_reserve_caps(struct ceph_mds_client *mdsc,
608 struct ceph_cap_reservation *ctx, int need);
609 extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
610 struct ceph_cap_reservation *ctx);
611 extern void ceph_reservation_status(struct ceph_fs_client *client,
612 int *total, int *avail, int *used,
613 int *reserved, int *min);
618 * we keep buffered readdir results attached to file->private_data
620 #define CEPH_F_SYNC 1
621 #define CEPH_F_ATEND 2
623 struct ceph_file_info {
624 short fmode; /* initialized on open */
625 short flags; /* CEPH_F_* */
627 /* readdir: position within the dir */
629 struct ceph_mds_request *last_readdir;
631 /* readdir: position within a frag */
632 unsigned next_offset; /* offset of next chunk (last_name's + 1) */
633 char *last_name; /* last entry in previous chunk */
634 long long dir_release_count;
635 long long dir_ordered_count;
636 int readdir_cache_idx;
638 /* used for -o dirstat read() on directory thing */
643 struct ceph_readdir_cache_control {
645 struct dentry **dentries;
650 * 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.
655 * All inodes within the realm that have capabilities are linked into a
658 struct ceph_snap_realm {
665 u64 parent_since; /* snapid when our current parent became so */
667 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
668 u32 num_prior_parent_snaps; /* had prior to parent_since */
669 u64 *snaps; /* snaps specific to this realm */
672 struct ceph_snap_realm *parent;
673 struct list_head children; /* list of child realms */
674 struct list_head child_item;
676 struct list_head empty_item; /* if i have ref==0 */
678 struct list_head dirty_item; /* if realm needs new context */
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;
687 static inline int default_congestion_kb(void)
694 * congestion size, scale with available memory.
706 * This allows larger machines to have larger/more transfers.
707 * Limit the default to 256M
709 congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
710 if (congestion_kb > 256*1024)
711 congestion_kb = 256*1024;
713 return congestion_kb;
719 struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
721 extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
722 struct ceph_snap_realm *realm);
723 extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
724 struct ceph_snap_realm *realm);
725 extern int ceph_update_snap_trace(struct ceph_mds_client *m,
726 void *p, void *e, bool deletion,
727 struct ceph_snap_realm **realm_ret);
728 extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
729 struct ceph_mds_session *session,
730 struct ceph_msg *msg);
731 extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
732 extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
733 struct ceph_cap_snap *capsnap);
734 extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
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.).
740 static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
742 return !list_empty(&ci->i_cap_snaps) &&
743 list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
748 extern const struct inode_operations ceph_file_iops;
750 extern struct inode *ceph_alloc_inode(struct super_block *sb);
751 extern void ceph_destroy_inode(struct inode *inode);
752 extern int ceph_drop_inode(struct inode *inode);
754 extern struct inode *ceph_get_inode(struct super_block *sb,
755 struct ceph_vino vino);
756 extern struct inode *ceph_get_snapdir(struct inode *parent);
757 extern int ceph_fill_file_size(struct inode *inode, int issued,
758 u32 truncate_seq, u64 truncate_size, u64 size);
759 extern 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);
762 extern int ceph_fill_trace(struct super_block *sb,
763 struct ceph_mds_request *req,
764 struct ceph_mds_session *session);
765 extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
766 struct ceph_mds_session *session);
768 extern int ceph_inode_holds_cap(struct inode *inode, int mask);
770 extern int ceph_inode_set_size(struct inode *inode, loff_t size);
771 extern void __ceph_do_pending_vmtruncate(struct inode *inode);
772 extern void ceph_queue_vmtruncate(struct inode *inode);
774 extern void ceph_queue_invalidate(struct inode *inode);
775 extern void ceph_queue_writeback(struct inode *inode);
777 extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
778 int mask, bool force);
779 static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
781 return __ceph_do_getattr(inode, NULL, mask, force);
783 extern int ceph_permission(struct inode *inode, int mask);
784 extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
785 extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
786 extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
790 int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
791 ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
792 extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
793 extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
794 extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
795 extern void __init ceph_xattr_init(void);
796 extern void ceph_xattr_exit(void);
797 extern const struct xattr_handler *ceph_xattr_handlers[];
799 #ifdef CONFIG_SECURITY
800 extern bool ceph_security_xattr_deadlock(struct inode *in);
801 extern bool ceph_security_xattr_wanted(struct inode *in);
803 static inline bool ceph_security_xattr_deadlock(struct inode *in)
807 static inline bool ceph_security_xattr_wanted(struct inode *in)
814 struct ceph_acls_info {
817 struct ceph_pagelist *pagelist;
820 #ifdef CONFIG_CEPH_FS_POSIX_ACL
822 struct posix_acl *ceph_get_acl(struct inode *, int);
823 int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type);
824 int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
825 struct ceph_acls_info *info);
826 void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info);
827 void ceph_release_acls_info(struct ceph_acls_info *info);
829 static inline void ceph_forget_all_cached_acls(struct inode *inode)
831 forget_all_cached_acls(inode);
836 #define ceph_get_acl NULL
837 #define ceph_set_acl NULL
839 static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
840 struct ceph_acls_info *info)
844 static inline void ceph_init_inode_acls(struct inode *inode,
845 struct ceph_acls_info *info)
848 static inline void ceph_release_acls_info(struct ceph_acls_info *info)
851 static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
856 static inline void ceph_forget_all_cached_acls(struct inode *inode)
863 extern const char *ceph_cap_string(int c);
864 extern void ceph_handle_caps(struct ceph_mds_session *session,
865 struct ceph_msg *msg);
866 extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
867 struct ceph_cap_reservation *ctx);
868 extern 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);
873 extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
874 extern void ceph_put_cap(struct ceph_mds_client *mdsc,
875 struct ceph_cap *cap);
876 extern int ceph_is_any_caps(struct inode *inode);
878 extern void ceph_queue_caps_release(struct inode *inode);
879 extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
880 extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
882 extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
883 struct ceph_mds_session *session);
884 extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
885 struct ceph_mds_session *session);
886 extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
888 extern int ceph_get_cap_mds(struct inode *inode);
889 extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
890 extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
891 extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
892 struct ceph_snap_context *snapc);
893 extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
894 struct ceph_mds_session **psession,
896 extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
897 struct ceph_mds_session *session);
898 extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
899 extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
901 extern int ceph_encode_inode_release(void **p, struct inode *inode,
902 int mds, int drop, int unless, int force);
903 extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
904 int mds, int drop, int unless);
906 extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
907 loff_t endoff, int *got, struct page **pinned_page);
909 /* for counting open files by mode */
910 static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
912 ci->i_nr_by_mode[mode]++;
914 extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
917 extern const struct address_space_operations ceph_aops;
918 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
919 extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
920 extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need);
921 extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
924 extern const struct file_operations ceph_file_fops;
926 extern int ceph_renew_caps(struct inode *inode);
927 extern int ceph_open(struct inode *inode, struct file *file);
928 extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
929 struct file *file, unsigned flags, umode_t mode,
931 extern int ceph_release(struct inode *inode, struct file *filp);
932 extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
933 char *data, size_t len);
935 extern const struct file_operations ceph_dir_fops;
936 extern const struct file_operations ceph_snapdir_fops;
937 extern const struct inode_operations ceph_dir_iops;
938 extern const struct inode_operations ceph_snapdir_iops;
939 extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
940 ceph_snapdir_dentry_ops;
942 extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
943 extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
944 extern int ceph_handle_snapdir(struct ceph_mds_request *req,
945 struct dentry *dentry, int err);
946 extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
947 struct dentry *dentry, int err);
949 extern void ceph_dentry_lru_add(struct dentry *dn);
950 extern void ceph_dentry_lru_touch(struct dentry *dn);
951 extern void ceph_dentry_lru_del(struct dentry *dn);
952 extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
953 extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
954 extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
957 * our d_ops vary depending on whether the inode is live,
958 * snapshotted (read-only), or a virtual ".snap" directory.
960 int ceph_init_dentry(struct dentry *dentry);
964 extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
967 extern const struct export_operations ceph_export_ops;
970 extern __init void ceph_flock_init(void);
971 extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
972 extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
973 extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
974 extern int ceph_encode_locks_to_buffer(struct inode *inode,
975 struct ceph_filelock *flocks,
977 int num_flock_locks);
978 extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
979 struct ceph_pagelist *pagelist,
980 int num_fcntl_locks, int num_flock_locks);
981 extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c);
984 extern int ceph_fs_debugfs_init(struct ceph_fs_client *client);
985 extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
987 #endif /* _FS_CEPH_SUPER_H */