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>
19 #include <linux/security.h>
21 #include <linux/ceph/libceph.h>
23 #ifdef CONFIG_CEPH_FSCACHE
24 #include <linux/fscache.h>
27 /* f_type in struct statfs */
28 #define CEPH_SUPER_MAGIC 0x00c36400
30 /* large granularity for statfs utilization stats to facilitate
31 * large volume sizes on 32-bit machines. */
32 #define CEPH_BLOCK_SHIFT 22 /* 4 MB */
33 #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT)
35 #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blacklisted */
36 #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */
37 #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */
38 #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */
39 #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */
40 #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */
41 #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */
42 #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */
43 #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */
44 #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */
45 #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */
46 #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */
48 #define CEPH_MOUNT_OPT_DEFAULT \
49 (CEPH_MOUNT_OPT_DCACHE | \
50 CEPH_MOUNT_OPT_NOCOPYFROM)
52 #define ceph_set_mount_opt(fsc, opt) \
53 (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt
54 #define ceph_clear_mount_opt(fsc, opt) \
55 (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt
56 #define ceph_test_mount_opt(fsc, opt) \
57 (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
59 /* max size of osd read request, limited by libceph */
60 #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN
61 /* osd has a configurable limitaion of max write size.
62 * CEPH_MSG_MAX_DATA_LEN should be small enough. */
63 #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN
64 #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */
65 #define CEPH_MAX_READDIR_DEFAULT 1024
66 #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024)
67 #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
70 * Delay telling the MDS we no longer want caps, in case we reopen
71 * the file. Delay a minimum amount of time, even if we send a cap
72 * message for some other reason. Otherwise, take the oppotunity to
73 * update the mds to avoid sending another message later.
75 #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */
76 #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */
78 struct ceph_mount_options {
81 unsigned int wsize; /* max write size */
82 unsigned int rsize; /* max read size */
83 unsigned int rasize; /* max readahead */
84 unsigned int congestion_kb; /* max writeback in flight */
85 unsigned int caps_wanted_delay_min, caps_wanted_delay_max;
87 unsigned int max_readdir; /* max readdir result (entries) */
88 unsigned int max_readdir_bytes; /* max readdir result (bytes) */
91 * everything above this point can be memcmp'd; everything below
92 * is handled in compare_mount_options()
95 char *snapdir_name; /* default ".snap" */
96 char *mds_namespace; /* default NULL */
97 char *server_path; /* default NULL (means "/") */
98 char *fscache_uniq; /* default NULL */
101 struct ceph_fs_client {
102 struct super_block *sb;
104 struct ceph_mount_options *mount_options;
105 struct ceph_client *client;
107 unsigned long mount_state;
109 unsigned long last_auto_reconnect;
112 bool have_copy_from2;
115 loff_t max_file_size;
117 struct ceph_mds_client *mdsc;
120 mempool_t *wb_pagevec_pool;
121 atomic_long_t writeback_count;
123 struct workqueue_struct *inode_wq;
124 struct workqueue_struct *cap_wq;
126 #ifdef CONFIG_DEBUG_FS
127 struct dentry *debugfs_dentry_lru, *debugfs_caps;
128 struct dentry *debugfs_congestion_kb;
129 struct dentry *debugfs_bdi;
130 struct dentry *debugfs_mdsc, *debugfs_mdsmap;
131 struct dentry *debugfs_mds_sessions;
134 #ifdef CONFIG_CEPH_FSCACHE
135 struct fscache_cookie *fscache;
141 * File i/o capability. This tracks shared state with the metadata
142 * server that allows us to cache or writeback attributes or to read
143 * and write data. For any given inode, we should have one or more
144 * capabilities, one issued by each metadata server, and our
145 * cumulative access is the OR of all issued capabilities.
147 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
148 * session capability lists.
151 struct ceph_inode_info *ci;
152 struct rb_node ci_node; /* per-ci cap tree */
153 struct ceph_mds_session *session;
154 struct list_head session_caps; /* per-session caplist */
155 u64 cap_id; /* unique cap id (mds provided) */
159 int issued; /* latest, from the mds */
160 int implemented; /* implemented superset of
161 issued (for revocation) */
164 /* caps to release */
170 u32 seq, issue_seq, mseq;
171 u32 cap_gen; /* active/stale cycle */
172 unsigned long last_used;
173 struct list_head caps_item;
176 #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */
177 #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */
178 #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */
180 struct ceph_cap_flush {
182 int caps; /* 0 means capsnap */
183 bool wake; /* wake up flush waiters when finish ? */
184 struct list_head g_list; // global
185 struct list_head i_list; // per inode
189 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
190 * we first complete any in-process sync writes and writeback any dirty
191 * data before flushing the snapped state (tracked here) back to the MDS.
193 struct ceph_cap_snap {
195 struct list_head ci_item;
197 struct ceph_cap_flush cap_flush;
201 struct ceph_snap_context *context;
207 struct ceph_buffer *xattr_blob;
212 struct timespec64 mtime, atime, ctime, btime;
216 int writing; /* a sync write is still in progress */
217 int dirty_pages; /* dirty pages awaiting writeback */
222 static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
224 if (refcount_dec_and_test(&capsnap->nref)) {
225 if (capsnap->xattr_blob)
226 ceph_buffer_put(capsnap->xattr_blob);
232 * The frag tree describes how a directory is fragmented, potentially across
233 * multiple metadata servers. It is also used to indicate points where
234 * metadata authority is delegated, and whether/where metadata is replicated.
236 * A _leaf_ frag will be present in the i_fragtree IFF there is
237 * delegation info. That is, if mds >= 0 || ndist > 0.
239 #define CEPH_MAX_DIRFRAG_REP 4
241 struct ceph_inode_frag {
246 int split_by; /* i.e. 2^(split_by) children */
248 /* delegation and replication info */
249 int mds; /* -1 if same authority as parent */
250 int ndist; /* >0 if replicated */
251 int dist[CEPH_MAX_DIRFRAG_REP];
255 * We cache inode xattrs as an encoded blob until they are first used,
256 * at which point we parse them into an rbtree.
258 struct ceph_inode_xattr {
267 int should_free_name;
274 struct ceph_dentry_info {
275 struct dentry *dentry;
276 struct ceph_mds_session *lease_session;
277 struct list_head lease_list;
279 int lease_shared_gen;
282 unsigned long lease_renew_after, lease_renew_from;
287 #define CEPH_DENTRY_REFERENCED 1
288 #define CEPH_DENTRY_LEASE_LIST 2
289 #define CEPH_DENTRY_SHRINK_LIST 4
290 #define CEPH_DENTRY_PRIMARY_LINK 8
292 struct ceph_inode_xattrs_info {
294 * (still encoded) xattr blob. we avoid the overhead of parsing
295 * this until someone actually calls getxattr, etc.
297 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
298 * NULL means we don't know.
300 struct ceph_buffer *blob, *prealloc_blob;
302 struct rb_root index;
307 u64 version, index_version;
313 struct ceph_inode_info {
314 struct ceph_vino i_vino; /* ceph ino + snap */
316 spinlock_t i_ceph_lock;
319 u64 i_inline_version;
322 unsigned long i_ceph_flags;
323 atomic64_t i_release_count;
324 atomic64_t i_ordered_count;
325 atomic64_t i_complete_seq[2];
327 struct ceph_dir_layout i_dir_layout;
328 struct ceph_file_layout i_layout;
329 struct ceph_file_layout i_cached_layout; // for async creates
333 struct timespec64 i_rctime;
334 u64 i_rbytes, i_rfiles, i_rsubdirs;
335 u64 i_files, i_subdirs;
338 u64 i_max_bytes, i_max_files;
342 struct rb_root i_fragtree;
343 int i_fragtree_nsplits;
344 struct mutex i_fragtree_mutex;
346 struct ceph_inode_xattrs_info i_xattrs;
348 /* capabilities. protected _both_ by i_ceph_lock and cap->session's
350 struct rb_root i_caps; /* cap list */
351 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
352 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
353 struct list_head i_dirty_item, i_flushing_item;
354 /* we need to track cap writeback on a per-cap-bit basis, to allow
355 * overlapping, pipelined cap flushes to the mds. we can probably
356 * reduce the tid to 8 bits if we're concerned about inode size. */
357 struct ceph_cap_flush *i_prealloc_cap_flush;
358 struct list_head i_cap_flush_list;
359 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
360 unsigned long i_hold_caps_max; /* jiffies */
361 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
362 struct ceph_cap_reservation i_cap_migration_resv;
363 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
364 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or
365 dirty|flushing caps */
366 unsigned i_snap_caps; /* cap bits for snapped files */
368 unsigned long i_last_rd;
369 unsigned long i_last_wr;
370 int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */
372 struct mutex i_truncate_mutex;
373 u32 i_truncate_seq; /* last truncate to smaller size */
374 u64 i_truncate_size; /* and the size we last truncated down to */
375 int i_truncate_pending; /* still need to call vmtruncate */
377 u64 i_max_size; /* max file size authorized by mds */
378 u64 i_reported_size; /* (max_)size reported to or requested of mds */
379 u64 i_wanted_max_size; /* offset we'd like to write too */
380 u64 i_requested_max_size; /* max_size we've requested */
382 /* held references to caps */
384 int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref;
385 int i_wrbuffer_ref, i_wrbuffer_ref_head;
386 atomic_t i_filelock_ref;
387 atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */
388 u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */
389 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
391 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
392 struct list_head i_unsafe_iops; /* uncommitted mds inode ops */
393 spinlock_t i_unsafe_lock;
396 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
397 struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */
399 int i_snap_realm_counter; /* snap realm (if caps) */
400 struct list_head i_snap_realm_item;
401 struct list_head i_snap_flush_item;
402 struct timespec64 i_btime;
403 struct timespec64 i_snap_btime;
405 struct work_struct i_work;
406 unsigned long i_work_mask;
408 #ifdef CONFIG_CEPH_FSCACHE
409 struct fscache_cookie *fscache;
414 struct inode vfs_inode; /* at end */
417 static inline struct ceph_inode_info *
418 ceph_inode(const struct inode *inode)
420 return container_of(inode, struct ceph_inode_info, vfs_inode);
423 static inline struct ceph_fs_client *
424 ceph_inode_to_client(const struct inode *inode)
426 return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
429 static inline struct ceph_fs_client *
430 ceph_sb_to_client(const struct super_block *sb)
432 return (struct ceph_fs_client *)sb->s_fs_info;
435 static inline struct ceph_vino
436 ceph_vino(const struct inode *inode)
438 return ceph_inode(inode)->i_vino;
442 * ino_t is <64 bits on many architectures, blech.
444 * i_ino (kernel inode) st_ino (userspace)
449 static inline u32 ceph_ino_to_ino32(__u64 vino)
451 u32 ino = vino & 0xffffffff;
461 static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
463 #if BITS_PER_LONG == 32
464 return ceph_ino_to_ino32(vino.ino);
466 return (ino_t)vino.ino;
471 * user-visible ino (stat, filldir)
473 #if BITS_PER_LONG == 32
474 static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
479 static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
481 if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))
482 ino = ceph_ino_to_ino32(ino);
488 /* for printf-style formatting */
489 #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
491 static inline u64 ceph_ino(struct inode *inode)
493 return ceph_inode(inode)->i_vino.ino;
495 static inline u64 ceph_snap(struct inode *inode)
497 return ceph_inode(inode)->i_vino.snap;
500 static inline int ceph_ino_compare(struct inode *inode, void *data)
502 struct ceph_vino *pvino = (struct ceph_vino *)data;
503 struct ceph_inode_info *ci = ceph_inode(inode);
504 return ci->i_vino.ino == pvino->ino &&
505 ci->i_vino.snap == pvino->snap;
508 static inline struct inode *ceph_find_inode(struct super_block *sb,
509 struct ceph_vino vino)
511 ino_t t = ceph_vino_to_ino(vino);
512 return ilookup5(sb, t, ceph_ino_compare, &vino);
519 #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */
520 #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */
521 #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */
522 #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */
523 #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */
524 #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */
525 #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */
526 #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */
527 #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */
528 #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */
529 #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */
530 #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */
531 #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT)
534 * Masks of ceph inode work.
536 #define CEPH_I_WORK_WRITEBACK 0 /* writeback */
537 #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */
538 #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */
541 * We set the ERROR_WRITE bit when we start seeing write errors on an inode
542 * and then clear it when they start succeeding. Note that we do a lockless
543 * check first, and only take the lock if it looks like it needs to be changed.
544 * The write submission code just takes this as a hint, so we're not too
545 * worried if a few slip through in either direction.
547 static inline void ceph_set_error_write(struct ceph_inode_info *ci)
549 if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) {
550 spin_lock(&ci->i_ceph_lock);
551 ci->i_ceph_flags |= CEPH_I_ERROR_WRITE;
552 spin_unlock(&ci->i_ceph_lock);
556 static inline void ceph_clear_error_write(struct ceph_inode_info *ci)
558 if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) {
559 spin_lock(&ci->i_ceph_lock);
560 ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE;
561 spin_unlock(&ci->i_ceph_lock);
565 static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
566 long long release_count,
567 long long ordered_count)
570 * Makes sure operations that setup readdir cache (update page
571 * cache and i_size) are strongly ordered w.r.t. the following
572 * atomic64_set() operations.
575 atomic64_set(&ci->i_complete_seq[0], release_count);
576 atomic64_set(&ci->i_complete_seq[1], ordered_count);
579 static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
581 atomic64_inc(&ci->i_release_count);
584 static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
586 atomic64_inc(&ci->i_ordered_count);
589 static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
591 return atomic64_read(&ci->i_complete_seq[0]) ==
592 atomic64_read(&ci->i_release_count);
595 static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
597 return atomic64_read(&ci->i_complete_seq[0]) ==
598 atomic64_read(&ci->i_release_count) &&
599 atomic64_read(&ci->i_complete_seq[1]) ==
600 atomic64_read(&ci->i_ordered_count);
603 static inline void ceph_dir_clear_complete(struct inode *inode)
605 __ceph_dir_clear_complete(ceph_inode(inode));
608 static inline void ceph_dir_clear_ordered(struct inode *inode)
610 __ceph_dir_clear_ordered(ceph_inode(inode));
613 static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
615 bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
620 /* find a specific frag @f */
621 extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
625 * choose fragment for value @v. copy frag content to pfrag, if leaf
628 extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
629 struct ceph_inode_frag *pfrag,
632 static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry)
634 return (struct ceph_dentry_info *)dentry->d_fsdata;
640 static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
642 return !RB_EMPTY_ROOT(&ci->i_caps);
645 extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
646 extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
647 extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
648 struct ceph_cap *cap);
650 static inline int ceph_caps_issued(struct ceph_inode_info *ci)
653 spin_lock(&ci->i_ceph_lock);
654 issued = __ceph_caps_issued(ci, NULL);
655 spin_unlock(&ci->i_ceph_lock);
659 static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
663 spin_lock(&ci->i_ceph_lock);
664 r = __ceph_caps_issued_mask(ci, mask, touch);
665 spin_unlock(&ci->i_ceph_lock);
669 static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
671 return ci->i_dirty_caps | ci->i_flushing_caps;
673 extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
674 extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
675 extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
676 struct ceph_cap_flush **pcf);
678 extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
679 struct ceph_cap *ocap, int mask);
680 extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
681 extern int __ceph_caps_used(struct ceph_inode_info *ci);
683 static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci)
685 return ci->i_nr_by_mode[0];
687 extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
688 extern int __ceph_caps_wanted(struct ceph_inode_info *ci);
690 /* what the mds thinks we want */
691 extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check);
693 extern void ceph_caps_init(struct ceph_mds_client *mdsc);
694 extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
695 extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc,
696 struct ceph_mount_options *fsopt);
697 extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
698 struct ceph_cap_reservation *ctx, int need);
699 extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc,
700 struct ceph_cap_reservation *ctx);
701 extern void ceph_reservation_status(struct ceph_fs_client *client,
702 int *total, int *avail, int *used,
703 int *reserved, int *min);
708 * we keep buffered readdir results attached to file->private_data
710 #define CEPH_F_SYNC 1
711 #define CEPH_F_ATEND 2
713 struct ceph_file_info {
714 short fmode; /* initialized on open */
715 short flags; /* CEPH_F_* */
717 spinlock_t rw_contexts_lock;
718 struct list_head rw_contexts;
725 struct ceph_dir_file_info {
726 struct ceph_file_info file_info;
728 /* readdir: position within the dir */
730 struct ceph_mds_request *last_readdir;
732 /* readdir: position within a frag */
733 unsigned next_offset; /* offset of next chunk (last_name's + 1) */
734 char *last_name; /* last entry in previous chunk */
735 long long dir_release_count;
736 long long dir_ordered_count;
737 int readdir_cache_idx;
739 /* used for -o dirstat read() on directory thing */
744 struct ceph_rw_context {
745 struct list_head list;
746 struct task_struct *thread;
750 #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \
751 struct ceph_rw_context _name = { \
756 static inline void ceph_add_rw_context(struct ceph_file_info *cf,
757 struct ceph_rw_context *ctx)
759 spin_lock(&cf->rw_contexts_lock);
760 list_add(&ctx->list, &cf->rw_contexts);
761 spin_unlock(&cf->rw_contexts_lock);
764 static inline void ceph_del_rw_context(struct ceph_file_info *cf,
765 struct ceph_rw_context *ctx)
767 spin_lock(&cf->rw_contexts_lock);
768 list_del(&ctx->list);
769 spin_unlock(&cf->rw_contexts_lock);
772 static inline struct ceph_rw_context*
773 ceph_find_rw_context(struct ceph_file_info *cf)
775 struct ceph_rw_context *ctx, *found = NULL;
776 spin_lock(&cf->rw_contexts_lock);
777 list_for_each_entry(ctx, &cf->rw_contexts, list) {
778 if (ctx->thread == current) {
783 spin_unlock(&cf->rw_contexts_lock);
787 struct ceph_readdir_cache_control {
789 struct dentry **dentries;
794 * A "snap realm" describes a subset of the file hierarchy sharing
795 * the same set of snapshots that apply to it. The realms themselves
796 * are organized into a hierarchy, such that children inherit (some of)
797 * the snapshots of their parents.
799 * All inodes within the realm that have capabilities are linked into a
802 struct ceph_snap_realm {
810 u64 parent_since; /* snapid when our current parent became so */
812 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
813 u32 num_prior_parent_snaps; /* had prior to parent_since */
814 u64 *snaps; /* snaps specific to this realm */
817 struct ceph_snap_realm *parent;
818 struct list_head children; /* list of child realms */
819 struct list_head child_item;
821 struct list_head empty_item; /* if i have ref==0 */
823 struct list_head dirty_item; /* if realm needs new context */
825 /* the current set of snaps for this realm */
826 struct ceph_snap_context *cached_context;
828 struct list_head inodes_with_caps;
829 spinlock_t inodes_with_caps_lock;
832 static inline int default_congestion_kb(void)
839 * congestion size, scale with available memory.
851 * This allows larger machines to have larger/more transfers.
852 * Limit the default to 256M
854 congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10);
855 if (congestion_kb > 256*1024)
856 congestion_kb = 256*1024;
858 return congestion_kb;
863 extern int ceph_force_reconnect(struct super_block *sb);
865 struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
867 extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
868 struct ceph_snap_realm *realm);
869 extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
870 struct ceph_snap_realm *realm);
871 extern int ceph_update_snap_trace(struct ceph_mds_client *m,
872 void *p, void *e, bool deletion,
873 struct ceph_snap_realm **realm_ret);
874 extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
875 struct ceph_mds_session *session,
876 struct ceph_msg *msg);
877 extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
878 extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
879 struct ceph_cap_snap *capsnap);
880 extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
882 extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc,
884 extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc,
885 struct ceph_snapid_map *sm);
886 extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc);
887 extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc);
891 * a cap_snap is "pending" if it is still awaiting an in-progress
892 * sync write (that may/may not still update size, mtime, etc.).
894 static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
896 return !list_empty(&ci->i_cap_snaps) &&
897 list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
902 struct ceph_mds_reply_info_in;
903 struct ceph_mds_reply_dirfrag;
905 extern const struct inode_operations ceph_file_iops;
907 extern struct inode *ceph_alloc_inode(struct super_block *sb);
908 extern void ceph_evict_inode(struct inode *inode);
909 extern void ceph_free_inode(struct inode *inode);
911 extern struct inode *ceph_get_inode(struct super_block *sb,
912 struct ceph_vino vino);
913 extern struct inode *ceph_get_snapdir(struct inode *parent);
914 extern int ceph_fill_file_size(struct inode *inode, int issued,
915 u32 truncate_seq, u64 truncate_size, u64 size);
916 extern void ceph_fill_file_time(struct inode *inode, int issued,
917 u64 time_warp_seq, struct timespec64 *ctime,
918 struct timespec64 *mtime,
919 struct timespec64 *atime);
920 extern int ceph_fill_inode(struct inode *inode, struct page *locked_page,
921 struct ceph_mds_reply_info_in *iinfo,
922 struct ceph_mds_reply_dirfrag *dirinfo,
923 struct ceph_mds_session *session, int cap_fmode,
924 struct ceph_cap_reservation *caps_reservation);
925 extern int ceph_fill_trace(struct super_block *sb,
926 struct ceph_mds_request *req);
927 extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
928 struct ceph_mds_session *session);
930 extern int ceph_inode_holds_cap(struct inode *inode, int mask);
932 extern bool ceph_inode_set_size(struct inode *inode, loff_t size);
933 extern void __ceph_do_pending_vmtruncate(struct inode *inode);
934 extern void ceph_queue_vmtruncate(struct inode *inode);
935 extern void ceph_queue_invalidate(struct inode *inode);
936 extern void ceph_queue_writeback(struct inode *inode);
937 extern void ceph_async_iput(struct inode *inode);
939 extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
940 int mask, bool force);
941 static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
943 return __ceph_do_getattr(inode, NULL, mask, force);
945 extern int ceph_permission(struct inode *inode, int mask);
946 extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
947 extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
948 extern int ceph_getattr(const struct path *path, struct kstat *stat,
949 u32 request_mask, unsigned int flags);
952 int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
953 ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
954 extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
955 extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci);
956 extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
957 extern const struct xattr_handler *ceph_xattr_handlers[];
959 struct ceph_acl_sec_ctx {
960 #ifdef CONFIG_CEPH_FS_POSIX_ACL
964 #ifdef CONFIG_CEPH_FS_SECURITY_LABEL
968 struct ceph_pagelist *pagelist;
971 #ifdef CONFIG_SECURITY
972 extern bool ceph_security_xattr_deadlock(struct inode *in);
973 extern bool ceph_security_xattr_wanted(struct inode *in);
975 static inline bool ceph_security_xattr_deadlock(struct inode *in)
979 static inline bool ceph_security_xattr_wanted(struct inode *in)
985 #ifdef CONFIG_CEPH_FS_SECURITY_LABEL
986 extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
987 struct ceph_acl_sec_ctx *ctx);
988 static inline void ceph_security_invalidate_secctx(struct inode *inode)
990 security_inode_invalidate_secctx(inode);
993 static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
994 struct ceph_acl_sec_ctx *ctx)
998 static inline void ceph_security_invalidate_secctx(struct inode *inode)
1003 void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx);
1006 #ifdef CONFIG_CEPH_FS_POSIX_ACL
1008 struct posix_acl *ceph_get_acl(struct inode *, int);
1009 int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type);
1010 int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
1011 struct ceph_acl_sec_ctx *as_ctx);
1012 void ceph_init_inode_acls(struct inode *inode,
1013 struct ceph_acl_sec_ctx *as_ctx);
1015 static inline void ceph_forget_all_cached_acls(struct inode *inode)
1017 forget_all_cached_acls(inode);
1022 #define ceph_get_acl NULL
1023 #define ceph_set_acl NULL
1025 static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
1026 struct ceph_acl_sec_ctx *as_ctx)
1030 static inline void ceph_init_inode_acls(struct inode *inode,
1031 struct ceph_acl_sec_ctx *as_ctx)
1034 static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
1039 static inline void ceph_forget_all_cached_acls(struct inode *inode)
1046 extern const char *ceph_cap_string(int c);
1047 extern void ceph_handle_caps(struct ceph_mds_session *session,
1048 struct ceph_msg *msg);
1049 extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
1050 struct ceph_cap_reservation *ctx);
1051 extern void ceph_add_cap(struct inode *inode,
1052 struct ceph_mds_session *session, u64 cap_id,
1053 unsigned issued, unsigned wanted,
1054 unsigned cap, unsigned seq, u64 realmino, int flags,
1055 struct ceph_cap **new_cap);
1056 extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
1057 extern void __ceph_remove_caps(struct ceph_inode_info *ci);
1058 extern void ceph_put_cap(struct ceph_mds_client *mdsc,
1059 struct ceph_cap *cap);
1060 extern int ceph_is_any_caps(struct inode *inode);
1062 extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
1063 extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
1065 extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
1066 struct ceph_mds_session *session);
1067 extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
1068 struct ceph_mds_session *session);
1069 void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session,
1070 struct ceph_inode_info *ci);
1071 extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
1073 extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps,
1074 bool snap_rwsem_locked);
1075 extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
1076 extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
1077 extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
1078 struct ceph_snap_context *snapc);
1079 extern void ceph_flush_snaps(struct ceph_inode_info *ci,
1080 struct ceph_mds_session **psession);
1081 extern bool __ceph_should_report_size(struct ceph_inode_info *ci);
1082 extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
1083 struct ceph_mds_session *session);
1084 extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
1085 extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
1086 extern int ceph_drop_caps_for_unlink(struct inode *inode);
1087 extern int ceph_encode_inode_release(void **p, struct inode *inode,
1088 int mds, int drop, int unless, int force);
1089 extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
1091 int mds, int drop, int unless);
1093 extern int ceph_get_caps(struct file *filp, int need, int want,
1094 loff_t endoff, int *got, struct page **pinned_page);
1095 extern int ceph_try_get_caps(struct inode *inode,
1096 int need, int want, bool nonblock, int *got);
1098 /* for counting open files by mode */
1099 extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count);
1100 extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count);
1101 extern void __ceph_touch_fmode(struct ceph_inode_info *ci,
1102 struct ceph_mds_client *mdsc, int fmode);
1105 extern const struct address_space_operations ceph_aops;
1106 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
1107 extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
1108 extern int ceph_pool_perm_check(struct inode *inode, int need);
1109 extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
1112 extern const struct file_operations ceph_file_fops;
1114 extern int ceph_renew_caps(struct inode *inode, int fmode);
1115 extern int ceph_open(struct inode *inode, struct file *file);
1116 extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
1117 struct file *file, unsigned flags, umode_t mode);
1118 extern int ceph_release(struct inode *inode, struct file *filp);
1119 extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
1120 char *data, size_t len);
1123 extern const struct file_operations ceph_dir_fops;
1124 extern const struct file_operations ceph_snapdir_fops;
1125 extern const struct inode_operations ceph_dir_iops;
1126 extern const struct inode_operations ceph_snapdir_iops;
1127 extern const struct dentry_operations ceph_dentry_ops;
1129 extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
1130 extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
1131 extern int ceph_handle_snapdir(struct ceph_mds_request *req,
1132 struct dentry *dentry, int err);
1133 extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
1134 struct dentry *dentry, int err);
1136 extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di);
1137 extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di);
1138 extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
1139 extern int ceph_trim_dentries(struct ceph_mds_client *mdsc);
1140 extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
1141 extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
1144 extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1147 extern const struct export_operations ceph_export_ops;
1148 struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino);
1151 extern __init void ceph_flock_init(void);
1152 extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
1153 extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
1154 extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
1155 extern int ceph_encode_locks_to_buffer(struct inode *inode,
1156 struct ceph_filelock *flocks,
1157 int num_fcntl_locks,
1158 int num_flock_locks);
1159 extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
1160 struct ceph_pagelist *pagelist,
1161 int num_fcntl_locks, int num_flock_locks);
1164 extern void ceph_fs_debugfs_init(struct ceph_fs_client *client);
1165 extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
1168 static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci)
1170 return ci->i_max_files || ci->i_max_bytes;
1173 extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc);
1175 static inline void __ceph_update_quota(struct ceph_inode_info *ci,
1176 u64 max_bytes, u64 max_files)
1178 bool had_quota, has_quota;
1179 had_quota = __ceph_has_any_quota(ci);
1180 ci->i_max_bytes = max_bytes;
1181 ci->i_max_files = max_files;
1182 has_quota = __ceph_has_any_quota(ci);
1184 if (had_quota != has_quota)
1185 ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota);
1188 extern void ceph_handle_quota(struct ceph_mds_client *mdsc,
1189 struct ceph_mds_session *session,
1190 struct ceph_msg *msg);
1191 extern bool ceph_quota_is_max_files_exceeded(struct inode *inode);
1192 extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new);
1193 extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode,
1195 extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode,
1197 extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc,
1198 struct kstatfs *buf);
1199 extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc);
1201 #endif /* _FS_CEPH_SUPER_H */