1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _FS_CEPH_SUPER_H
3 #define _FS_CEPH_SUPER_H
5 #include <linux/ceph/ceph_debug.h>
7 #include <asm/unaligned.h>
8 #include <linux/backing-dev.h>
9 #include <linux/completion.h>
10 #include <linux/exportfs.h>
12 #include <linux/mempool.h>
13 #include <linux/pagemap.h>
14 #include <linux/wait.h>
15 #include <linux/writeback.h>
16 #include <linux/slab.h>
17 #include <linux/posix_acl.h>
18 #include <linux/refcount.h>
20 #include <linux/ceph/libceph.h>
22 #ifdef CONFIG_CEPH_FSCACHE
23 #include <linux/fscache.h>
26 /* f_type in struct statfs */
27 #define CEPH_SUPER_MAGIC 0x00c36400
29 /* large granularity for statfs utilization stats to facilitate
30 * large volume sizes on 32-bit machines. */
31 #define CEPH_BLOCK_SHIFT 22 /* 4 MB */
32 #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT)
34 #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */
35 #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */
36 #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */
37 #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */
38 #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */
39 #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */
40 #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */
41 #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */
42 #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */
44 #define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE
46 #define ceph_set_mount_opt(fsc, opt) \
47 (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt;
48 #define ceph_test_mount_opt(fsc, opt) \
49 (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
51 /* max size of osd read request, limited by libceph */
52 #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN
53 /* osd has a configurable limitaion of max write size.
54 * CEPH_MSG_MAX_DATA_LEN should be small enough. */
55 #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN
56 #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */
57 #define CEPH_MAX_READDIR_DEFAULT 1024
58 #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024)
59 #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
62 * Delay telling the MDS we no longer want caps, in case we reopen
63 * the file. Delay a minimum amount of time, even if we send a cap
64 * message for some other reason. Otherwise, take the oppotunity to
65 * update the mds to avoid sending another message later.
67 #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */
68 #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */
70 struct ceph_mount_options {
74 int wsize; /* max write size */
75 int rsize; /* max read size */
76 int rasize; /* max readahead */
77 int congestion_kb; /* max writeback in flight */
78 int caps_wanted_delay_min, caps_wanted_delay_max;
79 int max_readdir; /* max readdir result (entires) */
80 int max_readdir_bytes; /* max readdir result (bytes) */
83 * everything above this point can be memcmp'd; everything below
84 * is handled in compare_mount_options()
87 char *snapdir_name; /* default ".snap" */
88 char *mds_namespace; /* default NULL */
89 char *server_path; /* default "/" */
90 char *fscache_uniq; /* default NULL */
93 struct ceph_fs_client {
94 struct super_block *sb;
96 struct ceph_mount_options *mount_options;
97 struct ceph_client *client;
99 unsigned long mount_state;
100 int min_caps; /* min caps i added */
102 struct ceph_mds_client *mdsc;
105 mempool_t *wb_pagevec_pool;
106 struct workqueue_struct *wb_wq;
107 struct workqueue_struct *pg_inv_wq;
108 struct workqueue_struct *trunc_wq;
109 atomic_long_t writeback_count;
111 #ifdef CONFIG_DEBUG_FS
112 struct dentry *debugfs_dentry_lru, *debugfs_caps;
113 struct dentry *debugfs_congestion_kb;
114 struct dentry *debugfs_bdi;
115 struct dentry *debugfs_mdsc, *debugfs_mdsmap;
116 struct dentry *debugfs_mds_sessions;
119 #ifdef CONFIG_CEPH_FSCACHE
120 struct fscache_cookie *fscache;
126 * File i/o capability. This tracks shared state with the metadata
127 * server that allows us to cache or writeback attributes or to read
128 * and write data. For any given inode, we should have one or more
129 * capabilities, one issued by each metadata server, and our
130 * cumulative access is the OR of all issued capabilities.
132 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
133 * session capability lists.
136 struct ceph_inode_info *ci;
137 struct rb_node ci_node; /* per-ci cap tree */
138 struct ceph_mds_session *session;
139 struct list_head session_caps; /* per-session caplist */
140 u64 cap_id; /* unique cap id (mds provided) */
144 int issued; /* latest, from the mds */
145 int implemented; /* implemented superset of
146 issued (for revocation) */
149 /* caps to release */
155 u32 seq, issue_seq, mseq;
156 u32 cap_gen; /* active/stale cycle */
157 unsigned long last_used;
158 struct list_head caps_item;
161 #define CHECK_CAPS_NODELAY 1 /* do not delay any further */
162 #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */
163 #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */
165 struct ceph_cap_flush {
167 int caps; /* 0 means capsnap */
168 bool wake; /* wake up flush waiters when finish ? */
169 struct list_head g_list; // global
170 struct list_head i_list; // per inode
174 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
175 * we first complete any in-process sync writes and writeback any dirty
176 * data before flushing the snapped state (tracked here) back to the MDS.
178 struct ceph_cap_snap {
180 struct list_head ci_item;
182 struct ceph_cap_flush cap_flush;
186 struct ceph_snap_context *context;
192 struct ceph_buffer *xattr_blob;
196 struct timespec mtime, atime, ctime;
200 int writing; /* a sync write is still in progress */
201 int dirty_pages; /* dirty pages awaiting writeback */
206 static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
208 if (refcount_dec_and_test(&capsnap->nref)) {
209 if (capsnap->xattr_blob)
210 ceph_buffer_put(capsnap->xattr_blob);
216 * The frag tree describes how a directory is fragmented, potentially across
217 * multiple metadata servers. It is also used to indicate points where
218 * metadata authority is delegated, and whether/where metadata is replicated.
220 * A _leaf_ frag will be present in the i_fragtree IFF there is
221 * delegation info. That is, if mds >= 0 || ndist > 0.
223 #define CEPH_MAX_DIRFRAG_REP 4
225 struct ceph_inode_frag {
230 int split_by; /* i.e. 2^(split_by) children */
232 /* delegation and replication info */
233 int mds; /* -1 if same authority as parent */
234 int ndist; /* >0 if replicated */
235 int dist[CEPH_MAX_DIRFRAG_REP];
239 * We cache inode xattrs as an encoded blob until they are first used,
240 * at which point we parse them into an rbtree.
242 struct ceph_inode_xattr {
251 int should_free_name;
258 struct ceph_dentry_info {
259 struct ceph_mds_session *lease_session;
260 int lease_shared_gen;
263 unsigned long lease_renew_after, lease_renew_from;
264 struct list_head lru;
265 struct dentry *dentry;
270 struct ceph_inode_xattrs_info {
272 * (still encoded) xattr blob. we avoid the overhead of parsing
273 * this until someone actually calls getxattr, etc.
275 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
276 * NULL means we don't know.
278 struct ceph_buffer *blob, *prealloc_blob;
280 struct rb_root index;
285 u64 version, index_version;
291 struct ceph_inode_info {
292 struct ceph_vino i_vino; /* ceph ino + snap */
294 spinlock_t i_ceph_lock;
297 u64 i_inline_version;
300 unsigned i_ceph_flags;
301 atomic64_t i_release_count;
302 atomic64_t i_ordered_count;
303 atomic64_t i_complete_seq[2];
305 struct ceph_dir_layout i_dir_layout;
306 struct ceph_file_layout i_layout;
310 struct timespec i_rctime;
311 u64 i_rbytes, i_rfiles, i_rsubdirs;
312 u64 i_files, i_subdirs;
315 u64 i_max_bytes, i_max_files;
317 struct rb_root i_fragtree;
318 int i_fragtree_nsplits;
319 struct mutex i_fragtree_mutex;
321 struct ceph_inode_xattrs_info i_xattrs;
323 /* capabilities. protected _both_ by i_ceph_lock and cap->session's
325 struct rb_root i_caps; /* cap list */
326 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
327 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
328 struct list_head i_dirty_item, i_flushing_item;
329 /* we need to track cap writeback on a per-cap-bit basis, to allow
330 * overlapping, pipelined cap flushes to the mds. we can probably
331 * reduce the tid to 8 bits if we're concerned about inode size. */
332 struct ceph_cap_flush *i_prealloc_cap_flush;
333 struct list_head i_cap_flush_list;
334 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
335 unsigned long i_hold_caps_min; /* jiffies */
336 unsigned long i_hold_caps_max; /* jiffies */
337 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
338 struct ceph_cap_reservation i_cap_migration_resv;
339 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
340 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or
341 dirty|flushing caps */
342 unsigned i_snap_caps; /* cap bits for snapped files */
344 int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */
346 struct mutex i_truncate_mutex;
347 u32 i_truncate_seq; /* last truncate to smaller size */
348 u64 i_truncate_size; /* and the size we last truncated down to */
349 int i_truncate_pending; /* still need to call vmtruncate */
351 u64 i_max_size; /* max file size authorized by mds */
352 u64 i_reported_size; /* (max_)size reported to or requested of mds */
353 u64 i_wanted_max_size; /* offset we'd like to write too */
354 u64 i_requested_max_size; /* max_size we've requested */
356 /* held references to caps */
358 int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref;
359 int i_wrbuffer_ref, i_wrbuffer_ref_head;
360 atomic_t i_filelock_ref;
361 atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */
362 u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */
363 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
365 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
366 struct list_head i_unsafe_iops; /* uncommitted mds inode ops */
367 spinlock_t i_unsafe_lock;
369 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
370 int i_snap_realm_counter; /* snap realm (if caps) */
371 struct list_head i_snap_realm_item;
372 struct list_head i_snap_flush_item;
374 struct work_struct i_wb_work; /* writeback work */
375 struct work_struct i_pg_inv_work; /* page invalidation work */
377 struct work_struct i_vmtruncate_work;
379 #ifdef CONFIG_CEPH_FSCACHE
380 struct fscache_cookie *fscache;
383 struct inode vfs_inode; /* at end */
386 static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
388 return container_of(inode, struct ceph_inode_info, vfs_inode);
391 static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode)
393 return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
396 static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb)
398 return (struct ceph_fs_client *)sb->s_fs_info;
401 static inline struct ceph_vino ceph_vino(struct inode *inode)
403 return ceph_inode(inode)->i_vino;
407 * ino_t is <64 bits on many architectures, blech.
409 * i_ino (kernel inode) st_ino (userspace)
414 static inline u32 ceph_ino_to_ino32(__u64 vino)
416 u32 ino = vino & 0xffffffff;
426 static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
428 #if BITS_PER_LONG == 32
429 return ceph_ino_to_ino32(vino.ino);
431 return (ino_t)vino.ino;
436 * user-visible ino (stat, filldir)
438 #if BITS_PER_LONG == 32
439 static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
444 static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
446 if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))
447 ino = ceph_ino_to_ino32(ino);
453 /* for printf-style formatting */
454 #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
456 static inline u64 ceph_ino(struct inode *inode)
458 return ceph_inode(inode)->i_vino.ino;
460 static inline u64 ceph_snap(struct inode *inode)
462 return ceph_inode(inode)->i_vino.snap;
465 static inline int ceph_ino_compare(struct inode *inode, void *data)
467 struct ceph_vino *pvino = (struct ceph_vino *)data;
468 struct ceph_inode_info *ci = ceph_inode(inode);
469 return ci->i_vino.ino == pvino->ino &&
470 ci->i_vino.snap == pvino->snap;
473 static inline struct inode *ceph_find_inode(struct super_block *sb,
474 struct ceph_vino vino)
476 ino_t t = ceph_vino_to_ino(vino);
477 return ilookup5(sb, t, ceph_ino_compare, &vino);
484 #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */
485 #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */
486 #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */
487 #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */
488 #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */
489 #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */
490 #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */
491 #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */
492 #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */
493 #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */
494 #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */
495 #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */
496 #define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */
500 * We set the ERROR_WRITE bit when we start seeing write errors on an inode
501 * and then clear it when they start succeeding. Note that we do a lockless
502 * check first, and only take the lock if it looks like it needs to be changed.
503 * The write submission code just takes this as a hint, so we're not too
504 * worried if a few slip through in either direction.
506 static inline void ceph_set_error_write(struct ceph_inode_info *ci)
508 if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) {
509 spin_lock(&ci->i_ceph_lock);
510 ci->i_ceph_flags |= CEPH_I_ERROR_WRITE;
511 spin_unlock(&ci->i_ceph_lock);
515 static inline void ceph_clear_error_write(struct ceph_inode_info *ci)
517 if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) {
518 spin_lock(&ci->i_ceph_lock);
519 ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE;
520 spin_unlock(&ci->i_ceph_lock);
524 static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
525 long long release_count,
526 long long ordered_count)
528 smp_mb__before_atomic();
529 atomic64_set(&ci->i_complete_seq[0], release_count);
530 atomic64_set(&ci->i_complete_seq[1], ordered_count);
533 static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
535 atomic64_inc(&ci->i_release_count);
538 static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
540 atomic64_inc(&ci->i_ordered_count);
543 static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
545 return atomic64_read(&ci->i_complete_seq[0]) ==
546 atomic64_read(&ci->i_release_count);
549 static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
551 return atomic64_read(&ci->i_complete_seq[0]) ==
552 atomic64_read(&ci->i_release_count) &&
553 atomic64_read(&ci->i_complete_seq[1]) ==
554 atomic64_read(&ci->i_ordered_count);
557 static inline void ceph_dir_clear_complete(struct inode *inode)
559 __ceph_dir_clear_complete(ceph_inode(inode));
562 static inline void ceph_dir_clear_ordered(struct inode *inode)
564 __ceph_dir_clear_ordered(ceph_inode(inode));
567 static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
569 bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
574 /* find a specific frag @f */
575 extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
579 * choose fragment for value @v. copy frag content to pfrag, if leaf
582 extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
583 struct ceph_inode_frag *pfrag,
586 static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
588 return (struct ceph_dentry_info *)dentry->d_fsdata;
594 static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
596 return !RB_EMPTY_ROOT(&ci->i_caps);
599 extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
600 extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
601 extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
602 struct ceph_cap *cap);
604 static inline int ceph_caps_issued(struct ceph_inode_info *ci)
607 spin_lock(&ci->i_ceph_lock);
608 issued = __ceph_caps_issued(ci, NULL);
609 spin_unlock(&ci->i_ceph_lock);
613 static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
617 spin_lock(&ci->i_ceph_lock);
618 r = __ceph_caps_issued_mask(ci, mask, touch);
619 spin_unlock(&ci->i_ceph_lock);
623 static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
625 return ci->i_dirty_caps | ci->i_flushing_caps;
627 extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
628 extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
629 extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
630 struct ceph_cap_flush **pcf);
632 extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
633 struct ceph_cap *ocap, int mask);
634 extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
635 extern int __ceph_caps_used(struct ceph_inode_info *ci);
637 extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
640 * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
642 static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
644 int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
645 if (w & CEPH_CAP_FILE_BUFFER)
646 w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */
650 /* what the mds thinks we want */
651 extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check);
653 extern void ceph_caps_init(struct ceph_mds_client *mdsc);
654 extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
655 extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
656 extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
657 struct ceph_cap_reservation *ctx, int need);
658 extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
659 struct ceph_cap_reservation *ctx);
660 extern void ceph_reservation_status(struct ceph_fs_client *client,
661 int *total, int *avail, int *used,
662 int *reserved, int *min);
667 * we keep buffered readdir results attached to file->private_data
669 #define CEPH_F_SYNC 1
670 #define CEPH_F_ATEND 2
672 struct ceph_file_info {
673 short fmode; /* initialized on open */
674 short flags; /* CEPH_F_* */
676 spinlock_t rw_contexts_lock;
677 struct list_head rw_contexts;
680 struct ceph_dir_file_info {
681 struct ceph_file_info file_info;
683 /* readdir: position within the dir */
685 struct ceph_mds_request *last_readdir;
687 /* readdir: position within a frag */
688 unsigned next_offset; /* offset of next chunk (last_name's + 1) */
689 char *last_name; /* last entry in previous chunk */
690 long long dir_release_count;
691 long long dir_ordered_count;
692 int readdir_cache_idx;
694 /* used for -o dirstat read() on directory thing */
699 struct ceph_rw_context {
700 struct list_head list;
701 struct task_struct *thread;
705 #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \
706 struct ceph_rw_context _name = { \
711 static inline void ceph_add_rw_context(struct ceph_file_info *cf,
712 struct ceph_rw_context *ctx)
714 spin_lock(&cf->rw_contexts_lock);
715 list_add(&ctx->list, &cf->rw_contexts);
716 spin_unlock(&cf->rw_contexts_lock);
719 static inline void ceph_del_rw_context(struct ceph_file_info *cf,
720 struct ceph_rw_context *ctx)
722 spin_lock(&cf->rw_contexts_lock);
723 list_del(&ctx->list);
724 spin_unlock(&cf->rw_contexts_lock);
727 static inline struct ceph_rw_context*
728 ceph_find_rw_context(struct ceph_file_info *cf)
730 struct ceph_rw_context *ctx, *found = NULL;
731 spin_lock(&cf->rw_contexts_lock);
732 list_for_each_entry(ctx, &cf->rw_contexts, list) {
733 if (ctx->thread == current) {
738 spin_unlock(&cf->rw_contexts_lock);
742 struct ceph_readdir_cache_control {
744 struct dentry **dentries;
749 * A "snap realm" describes a subset of the file hierarchy sharing
750 * the same set of snapshots that apply to it. The realms themselves
751 * are organized into a hierarchy, such that children inherit (some of)
752 * the snapshots of their parents.
754 * All inodes within the realm that have capabilities are linked into a
757 struct ceph_snap_realm {
765 u64 parent_since; /* snapid when our current parent became so */
767 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
768 u32 num_prior_parent_snaps; /* had prior to parent_since */
769 u64 *snaps; /* snaps specific to this realm */
772 struct ceph_snap_realm *parent;
773 struct list_head children; /* list of child realms */
774 struct list_head child_item;
776 struct list_head empty_item; /* if i have ref==0 */
778 struct list_head dirty_item; /* if realm needs new context */
780 /* the current set of snaps for this realm */
781 struct ceph_snap_context *cached_context;
783 struct list_head inodes_with_caps;
784 spinlock_t inodes_with_caps_lock;
787 static inline int default_congestion_kb(void)
794 * congestion size, scale with available memory.
806 * This allows larger machines to have larger/more transfers.
807 * Limit the default to 256M
809 congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
810 if (congestion_kb > 256*1024)
811 congestion_kb = 256*1024;
813 return congestion_kb;
819 struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
821 extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
822 struct ceph_snap_realm *realm);
823 extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
824 struct ceph_snap_realm *realm);
825 extern int ceph_update_snap_trace(struct ceph_mds_client *m,
826 void *p, void *e, bool deletion,
827 struct ceph_snap_realm **realm_ret);
828 extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
829 struct ceph_mds_session *session,
830 struct ceph_msg *msg);
831 extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
832 extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
833 struct ceph_cap_snap *capsnap);
834 extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
837 * a cap_snap is "pending" if it is still awaiting an in-progress
838 * sync write (that may/may not still update size, mtime, etc.).
840 static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
842 return !list_empty(&ci->i_cap_snaps) &&
843 list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
848 extern const struct inode_operations ceph_file_iops;
850 extern struct inode *ceph_alloc_inode(struct super_block *sb);
851 extern void ceph_destroy_inode(struct inode *inode);
852 extern int ceph_drop_inode(struct inode *inode);
854 extern struct inode *ceph_get_inode(struct super_block *sb,
855 struct ceph_vino vino);
856 extern struct inode *ceph_get_snapdir(struct inode *parent);
857 extern int ceph_fill_file_size(struct inode *inode, int issued,
858 u32 truncate_seq, u64 truncate_size, u64 size);
859 extern void ceph_fill_file_time(struct inode *inode, int issued,
860 u64 time_warp_seq, struct timespec *ctime,
861 struct timespec *mtime, struct timespec *atime);
862 extern int ceph_fill_trace(struct super_block *sb,
863 struct ceph_mds_request *req);
864 extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
865 struct ceph_mds_session *session);
867 extern int ceph_inode_holds_cap(struct inode *inode, int mask);
869 extern bool ceph_inode_set_size(struct inode *inode, loff_t size);
870 extern void __ceph_do_pending_vmtruncate(struct inode *inode);
871 extern void ceph_queue_vmtruncate(struct inode *inode);
873 extern void ceph_queue_invalidate(struct inode *inode);
874 extern void ceph_queue_writeback(struct inode *inode);
876 extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
877 int mask, bool force);
878 static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
880 return __ceph_do_getattr(inode, NULL, mask, force);
882 extern int ceph_permission(struct inode *inode, int mask);
883 extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
884 extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
885 extern int ceph_getattr(const struct path *path, struct kstat *stat,
886 u32 request_mask, unsigned int flags);
889 int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
890 ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
891 extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
892 extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
893 extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
894 extern void __init ceph_xattr_init(void);
895 extern void ceph_xattr_exit(void);
896 extern const struct xattr_handler *ceph_xattr_handlers[];
898 #ifdef CONFIG_SECURITY
899 extern bool ceph_security_xattr_deadlock(struct inode *in);
900 extern bool ceph_security_xattr_wanted(struct inode *in);
902 static inline bool ceph_security_xattr_deadlock(struct inode *in)
906 static inline bool ceph_security_xattr_wanted(struct inode *in)
913 struct ceph_acls_info {
916 struct ceph_pagelist *pagelist;
919 #ifdef CONFIG_CEPH_FS_POSIX_ACL
921 struct posix_acl *ceph_get_acl(struct inode *, int);
922 int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type);
923 int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
924 struct ceph_acls_info *info);
925 void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info);
926 void ceph_release_acls_info(struct ceph_acls_info *info);
928 static inline void ceph_forget_all_cached_acls(struct inode *inode)
930 forget_all_cached_acls(inode);
935 #define ceph_get_acl NULL
936 #define ceph_set_acl NULL
938 static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
939 struct ceph_acls_info *info)
943 static inline void ceph_init_inode_acls(struct inode *inode,
944 struct ceph_acls_info *info)
947 static inline void ceph_release_acls_info(struct ceph_acls_info *info)
950 static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
955 static inline void ceph_forget_all_cached_acls(struct inode *inode)
962 extern const char *ceph_cap_string(int c);
963 extern void ceph_handle_caps(struct ceph_mds_session *session,
964 struct ceph_msg *msg);
965 extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
966 struct ceph_cap_reservation *ctx);
967 extern void ceph_add_cap(struct inode *inode,
968 struct ceph_mds_session *session, u64 cap_id,
969 int fmode, unsigned issued, unsigned wanted,
970 unsigned cap, unsigned seq, u64 realmino, int flags,
971 struct ceph_cap **new_cap);
972 extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
973 extern void ceph_put_cap(struct ceph_mds_client *mdsc,
974 struct ceph_cap *cap);
975 extern int ceph_is_any_caps(struct inode *inode);
977 extern void ceph_queue_caps_release(struct inode *inode);
978 extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
979 extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
981 extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
982 struct ceph_mds_session *session);
983 extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
984 struct ceph_mds_session *session);
985 extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
987 extern int ceph_get_cap_mds(struct inode *inode);
988 extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
989 extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
990 extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
991 struct ceph_snap_context *snapc);
992 extern void ceph_flush_snaps(struct ceph_inode_info *ci,
993 struct ceph_mds_session **psession);
994 extern bool __ceph_should_report_size(struct ceph_inode_info *ci);
995 extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
996 struct ceph_mds_session *session);
997 extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
998 extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
999 extern int ceph_drop_caps_for_unlink(struct inode *inode);
1000 extern int ceph_encode_inode_release(void **p, struct inode *inode,
1001 int mds, int drop, int unless, int force);
1002 extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
1004 int mds, int drop, int unless);
1006 extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
1007 loff_t endoff, int *got, struct page **pinned_page);
1008 extern int ceph_try_get_caps(struct ceph_inode_info *ci,
1009 int need, int want, int *got);
1011 /* for counting open files by mode */
1012 extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode);
1013 extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
1016 extern const struct address_space_operations ceph_aops;
1017 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
1018 extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
1019 extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need);
1020 extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
1023 extern const struct file_operations ceph_file_fops;
1025 extern int ceph_renew_caps(struct inode *inode);
1026 extern int ceph_open(struct inode *inode, struct file *file);
1027 extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
1028 struct file *file, unsigned flags, umode_t mode);
1029 extern int ceph_release(struct inode *inode, struct file *filp);
1030 extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
1031 char *data, size_t len);
1034 extern const struct file_operations ceph_dir_fops;
1035 extern const struct file_operations ceph_snapdir_fops;
1036 extern const struct inode_operations ceph_dir_iops;
1037 extern const struct inode_operations ceph_snapdir_iops;
1038 extern const struct dentry_operations ceph_dentry_ops;
1040 extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
1041 extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
1042 extern int ceph_handle_snapdir(struct ceph_mds_request *req,
1043 struct dentry *dentry, int err);
1044 extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
1045 struct dentry *dentry, int err);
1047 extern void ceph_dentry_lru_add(struct dentry *dn);
1048 extern void ceph_dentry_lru_touch(struct dentry *dn);
1049 extern void ceph_dentry_lru_del(struct dentry *dn);
1050 extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
1051 extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
1052 extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
1055 extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1058 extern const struct export_operations ceph_export_ops;
1061 extern __init void ceph_flock_init(void);
1062 extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
1063 extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
1064 extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
1065 extern int ceph_encode_locks_to_buffer(struct inode *inode,
1066 struct ceph_filelock *flocks,
1067 int num_fcntl_locks,
1068 int num_flock_locks);
1069 extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
1070 struct ceph_pagelist *pagelist,
1071 int num_fcntl_locks, int num_flock_locks);
1074 extern int ceph_fs_debugfs_init(struct ceph_fs_client *client);
1075 extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
1078 static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci)
1080 return ci->i_max_files || ci->i_max_bytes;
1083 extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc);
1085 static inline void __ceph_update_quota(struct ceph_inode_info *ci,
1086 u64 max_bytes, u64 max_files)
1088 bool had_quota, has_quota;
1089 had_quota = __ceph_has_any_quota(ci);
1090 ci->i_max_bytes = max_bytes;
1091 ci->i_max_files = max_files;
1092 has_quota = __ceph_has_any_quota(ci);
1094 if (had_quota != has_quota)
1095 ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota);
1098 extern void ceph_handle_quota(struct ceph_mds_client *mdsc,
1099 struct ceph_mds_session *session,
1100 struct ceph_msg *msg);
1101 extern bool ceph_quota_is_max_files_exceeded(struct inode *inode);
1102 extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new);
1103 extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode,
1105 extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode,
1107 extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc,
1108 struct kstatfs *buf);
1110 #endif /* _FS_CEPH_SUPER_H */