xref: /linux/fs/ceph/super.h (revision cd045cb42a266882ac24bc21a3a8d03683c72954)
1de57606cSSage Weil #ifndef _FS_CEPH_SUPER_H
2de57606cSSage Weil #define _FS_CEPH_SUPER_H
3de57606cSSage Weil 
43d14c5d2SYehuda Sadeh #include <linux/ceph/ceph_debug.h>
5de57606cSSage Weil 
6de57606cSSage Weil #include <asm/unaligned.h>
7de57606cSSage Weil #include <linux/backing-dev.h>
8de57606cSSage Weil #include <linux/completion.h>
9de57606cSSage Weil #include <linux/exportfs.h>
10de57606cSSage Weil #include <linux/fs.h>
11de57606cSSage Weil #include <linux/mempool.h>
12de57606cSSage Weil #include <linux/pagemap.h>
13de57606cSSage Weil #include <linux/wait.h>
14f1a3d572SStephen Rothwell #include <linux/writeback.h>
155a0e3ad6STejun Heo #include <linux/slab.h>
16de57606cSSage Weil 
173d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h>
18de57606cSSage Weil 
19de57606cSSage Weil /* f_type in struct statfs */
20de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400
21de57606cSSage Weil 
22de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate
23de57606cSSage Weil  * large volume sizes on 32-bit machines. */
24de57606cSSage Weil #define CEPH_BLOCK_SHIFT   20  /* 1 MB */
25de57606cSSage Weil #define CEPH_BLOCK         (1 << CEPH_BLOCK_SHIFT)
26de57606cSSage Weil 
273d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT         (1<<4) /* `cat dirname` for stats */
283d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES          (1<<5) /* dir st_bytes = rbytes */
293d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR  (1<<7) /* no dcache readdir */
306a259382SSage Weil 
313d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DEFAULT    (CEPH_MOUNT_OPT_RBYTES)
32de57606cSSage Weil 
333d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \
343d14c5d2SYehuda Sadeh 	(fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt;
353d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \
363d14c5d2SYehuda Sadeh 	(!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
37de57606cSSage Weil 
383d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT        1024
393d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT  (512*1024)
403d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT        ".snap"
41de57606cSSage Weil 
423d14c5d2SYehuda Sadeh struct ceph_mount_options {
43de57606cSSage Weil 	int flags;
443d14c5d2SYehuda Sadeh 	int sb_flags;
453d14c5d2SYehuda Sadeh 
466e19a16eSSage Weil 	int wsize;
476e19a16eSSage Weil 	int rsize;            /* max readahead */
486e19a16eSSage Weil 	int congestion_kb;    /* max writeback in flight */
496e19a16eSSage Weil 	int caps_wanted_delay_min, caps_wanted_delay_max;
506e19a16eSSage Weil 	int cap_release_safety;
5123804d91SSage Weil 	int max_readdir;       /* max readdir result (entires) */
5223804d91SSage Weil 	int max_readdir_bytes; /* max readdir result (bytes) */
533d14c5d2SYehuda Sadeh 
543d14c5d2SYehuda Sadeh 	/*
553d14c5d2SYehuda Sadeh 	 * everything above this point can be memcmp'd; everything below
563d14c5d2SYehuda Sadeh 	 * is handled in compare_mount_options()
573d14c5d2SYehuda Sadeh 	 */
583d14c5d2SYehuda Sadeh 
59de57606cSSage Weil 	char *snapdir_name;   /* default ".snap" */
60de57606cSSage Weil };
61de57606cSSage Weil 
623d14c5d2SYehuda Sadeh struct ceph_fs_client {
63de57606cSSage Weil 	struct super_block *sb;
64de57606cSSage Weil 
653d14c5d2SYehuda Sadeh 	struct ceph_mount_options *mount_options;
663d14c5d2SYehuda Sadeh 	struct ceph_client *client;
673d14c5d2SYehuda Sadeh 
68de57606cSSage Weil 	unsigned long mount_state;
6985ccce43SSage Weil 	int min_caps;                  /* min caps i added */
7085ccce43SSage Weil 
713d14c5d2SYehuda Sadeh 	struct ceph_mds_client *mdsc;
72de57606cSSage Weil 
73de57606cSSage Weil 	/* writeback */
74de57606cSSage Weil 	mempool_t *wb_pagevec_pool;
75de57606cSSage Weil 	struct workqueue_struct *wb_wq;
76de57606cSSage Weil 	struct workqueue_struct *pg_inv_wq;
77de57606cSSage Weil 	struct workqueue_struct *trunc_wq;
782baba250SYehuda Sadeh 	atomic_long_t writeback_count;
79de57606cSSage Weil 
80de57606cSSage Weil 	struct backing_dev_info backing_dev_info;
810743304dSSage Weil 
820743304dSSage Weil #ifdef CONFIG_DEBUG_FS
833d14c5d2SYehuda Sadeh 	struct dentry *debugfs_dentry_lru, *debugfs_caps;
842baba250SYehuda Sadeh 	struct dentry *debugfs_congestion_kb;
8506edf046SSage Weil 	struct dentry *debugfs_bdi;
863d14c5d2SYehuda Sadeh 	struct dentry *debugfs_mdsc, *debugfs_mdsmap;
870743304dSSage Weil #endif
88de57606cSSage Weil };
89de57606cSSage Weil 
903d14c5d2SYehuda Sadeh 
91de57606cSSage Weil /*
92de57606cSSage Weil  * File i/o capability.  This tracks shared state with the metadata
93de57606cSSage Weil  * server that allows us to cache or writeback attributes or to read
94de57606cSSage Weil  * and write data.  For any given inode, we should have one or more
95de57606cSSage Weil  * capabilities, one issued by each metadata server, and our
96de57606cSSage Weil  * cumulative access is the OR of all issued capabilities.
97de57606cSSage Weil  *
98de57606cSSage Weil  * Each cap is referenced by the inode's i_caps rbtree and by per-mds
99de57606cSSage Weil  * session capability lists.
100de57606cSSage Weil  */
101de57606cSSage Weil struct ceph_cap {
102de57606cSSage Weil 	struct ceph_inode_info *ci;
103de57606cSSage Weil 	struct rb_node ci_node;          /* per-ci cap tree */
104de57606cSSage Weil 	struct ceph_mds_session *session;
105de57606cSSage Weil 	struct list_head session_caps;   /* per-session caplist */
106de57606cSSage Weil 	int mds;
107de57606cSSage Weil 	u64 cap_id;       /* unique cap id (mds provided) */
108de57606cSSage Weil 	int issued;       /* latest, from the mds */
109de57606cSSage Weil 	int implemented;  /* implemented superset of issued (for revocation) */
110de57606cSSage Weil 	int mds_wanted;
111685f9a5dSSage Weil 	u32 seq, issue_seq, mseq;
112685f9a5dSSage Weil 	u32 cap_gen;      /* active/stale cycle */
113de57606cSSage Weil 	unsigned long last_used;
114de57606cSSage Weil 	struct list_head caps_item;
115de57606cSSage Weil };
116de57606cSSage Weil 
117de57606cSSage Weil #define CHECK_CAPS_NODELAY    1  /* do not delay any further */
118de57606cSSage Weil #define CHECK_CAPS_AUTHONLY   2  /* only check auth cap */
119de57606cSSage Weil #define CHECK_CAPS_FLUSH      4  /* flush any dirty caps */
120de57606cSSage Weil 
121de57606cSSage Weil /*
122de57606cSSage Weil  * Snapped cap state that is pending flush to mds.  When a snapshot occurs,
123de57606cSSage Weil  * we first complete any in-process sync writes and writeback any dirty
124de57606cSSage Weil  * data before flushing the snapped state (tracked here) back to the MDS.
125de57606cSSage Weil  */
126de57606cSSage Weil struct ceph_cap_snap {
127de57606cSSage Weil 	atomic_t nref;
128de57606cSSage Weil 	struct ceph_inode_info *ci;
129de57606cSSage Weil 	struct list_head ci_item, flushing_item;
130de57606cSSage Weil 
131de57606cSSage Weil 	u64 follows, flush_tid;
132de57606cSSage Weil 	int issued, dirty;
133de57606cSSage Weil 	struct ceph_snap_context *context;
134de57606cSSage Weil 
135de57606cSSage Weil 	mode_t mode;
136de57606cSSage Weil 	uid_t uid;
137de57606cSSage Weil 	gid_t gid;
138de57606cSSage Weil 
1394a625be4SSage Weil 	struct ceph_buffer *xattr_blob;
140de57606cSSage Weil 	u64 xattr_version;
141de57606cSSage Weil 
142de57606cSSage Weil 	u64 size;
143de57606cSSage Weil 	struct timespec mtime, atime, ctime;
144de57606cSSage Weil 	u64 time_warp_seq;
145de57606cSSage Weil 	int writing;   /* a sync write is still in progress */
146de57606cSSage Weil 	int dirty_pages;     /* dirty pages awaiting writeback */
147de57606cSSage Weil };
148de57606cSSage Weil 
149de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
150de57606cSSage Weil {
1514a625be4SSage Weil 	if (atomic_dec_and_test(&capsnap->nref)) {
1524a625be4SSage Weil 		if (capsnap->xattr_blob)
1534a625be4SSage Weil 			ceph_buffer_put(capsnap->xattr_blob);
154de57606cSSage Weil 		kfree(capsnap);
155de57606cSSage Weil 	}
1564a625be4SSage Weil }
157de57606cSSage Weil 
158de57606cSSage Weil /*
159de57606cSSage Weil  * The frag tree describes how a directory is fragmented, potentially across
160de57606cSSage Weil  * multiple metadata servers.  It is also used to indicate points where
161de57606cSSage Weil  * metadata authority is delegated, and whether/where metadata is replicated.
162de57606cSSage Weil  *
163de57606cSSage Weil  * A _leaf_ frag will be present in the i_fragtree IFF there is
164de57606cSSage Weil  * delegation info.  That is, if mds >= 0 || ndist > 0.
165de57606cSSage Weil  */
166de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4
167de57606cSSage Weil 
168de57606cSSage Weil struct ceph_inode_frag {
169de57606cSSage Weil 	struct rb_node node;
170de57606cSSage Weil 
171de57606cSSage Weil 	/* fragtree state */
172de57606cSSage Weil 	u32 frag;
173de57606cSSage Weil 	int split_by;         /* i.e. 2^(split_by) children */
174de57606cSSage Weil 
175de57606cSSage Weil 	/* delegation and replication info */
176de57606cSSage Weil 	int mds;              /* -1 if same authority as parent */
177de57606cSSage Weil 	int ndist;            /* >0 if replicated */
178de57606cSSage Weil 	int dist[CEPH_MAX_DIRFRAG_REP];
179de57606cSSage Weil };
180de57606cSSage Weil 
181de57606cSSage Weil /*
182de57606cSSage Weil  * We cache inode xattrs as an encoded blob until they are first used,
183de57606cSSage Weil  * at which point we parse them into an rbtree.
184de57606cSSage Weil  */
185de57606cSSage Weil struct ceph_inode_xattr {
186de57606cSSage Weil 	struct rb_node node;
187de57606cSSage Weil 
188de57606cSSage Weil 	const char *name;
189de57606cSSage Weil 	int name_len;
190de57606cSSage Weil 	const char *val;
191de57606cSSage Weil 	int val_len;
192de57606cSSage Weil 	int dirty;
193de57606cSSage Weil 
194de57606cSSage Weil 	int should_free_name;
195de57606cSSage Weil 	int should_free_val;
196de57606cSSage Weil };
197de57606cSSage Weil 
1983d14c5d2SYehuda Sadeh /*
1993d14c5d2SYehuda Sadeh  * Ceph dentry state
2003d14c5d2SYehuda Sadeh  */
2013d14c5d2SYehuda Sadeh struct ceph_dentry_info {
2023d14c5d2SYehuda Sadeh 	struct ceph_mds_session *lease_session;
2033d14c5d2SYehuda Sadeh 	u32 lease_gen, lease_shared_gen;
2043d14c5d2SYehuda Sadeh 	u32 lease_seq;
2053d14c5d2SYehuda Sadeh 	unsigned long lease_renew_after, lease_renew_from;
2063d14c5d2SYehuda Sadeh 	struct list_head lru;
2073d14c5d2SYehuda Sadeh 	struct dentry *dentry;
2083d14c5d2SYehuda Sadeh 	u64 time;
2093d14c5d2SYehuda Sadeh 	u64 offset;
2103d14c5d2SYehuda Sadeh };
2113d14c5d2SYehuda Sadeh 
212de57606cSSage Weil struct ceph_inode_xattrs_info {
213de57606cSSage Weil 	/*
214de57606cSSage Weil 	 * (still encoded) xattr blob. we avoid the overhead of parsing
215de57606cSSage Weil 	 * this until someone actually calls getxattr, etc.
216de57606cSSage Weil 	 *
217de57606cSSage Weil 	 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
218de57606cSSage Weil 	 * NULL means we don't know.
219de57606cSSage Weil 	*/
220de57606cSSage Weil 	struct ceph_buffer *blob, *prealloc_blob;
221de57606cSSage Weil 
222de57606cSSage Weil 	struct rb_root index;
223de57606cSSage Weil 	bool dirty;
224de57606cSSage Weil 	int count;
225de57606cSSage Weil 	int names_size;
226de57606cSSage Weil 	int vals_size;
227de57606cSSage Weil 	u64 version, index_version;
228de57606cSSage Weil };
229de57606cSSage Weil 
230de57606cSSage Weil /*
231de57606cSSage Weil  * Ceph inode.
232de57606cSSage Weil  */
233de57606cSSage Weil struct ceph_inode_info {
234de57606cSSage Weil 	struct ceph_vino i_vino;   /* ceph ino + snap */
235de57606cSSage Weil 
236de57606cSSage Weil 	u64 i_version;
237de57606cSSage Weil 	u32 i_time_warp_seq;
238de57606cSSage Weil 
239de57606cSSage Weil 	unsigned i_ceph_flags;
240de57606cSSage Weil 	unsigned long i_release_count;
241de57606cSSage Weil 
242de57606cSSage Weil 	struct ceph_file_layout i_layout;
243de57606cSSage Weil 	char *i_symlink;
244de57606cSSage Weil 
245de57606cSSage Weil 	/* for dirs */
246de57606cSSage Weil 	struct timespec i_rctime;
247de57606cSSage Weil 	u64 i_rbytes, i_rfiles, i_rsubdirs;
248de57606cSSage Weil 	u64 i_files, i_subdirs;
249de57606cSSage Weil 	u64 i_max_offset;  /* largest readdir offset, set with I_COMPLETE */
250de57606cSSage Weil 
251de57606cSSage Weil 	struct rb_root i_fragtree;
252de57606cSSage Weil 	struct mutex i_fragtree_mutex;
253de57606cSSage Weil 
254de57606cSSage Weil 	struct ceph_inode_xattrs_info i_xattrs;
255de57606cSSage Weil 
256de57606cSSage Weil 	/* capabilities.  protected _both_ by i_lock and cap->session's
257de57606cSSage Weil 	 * s_mutex. */
258de57606cSSage Weil 	struct rb_root i_caps;           /* cap list */
259de57606cSSage Weil 	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
260de57606cSSage Weil 	unsigned i_dirty_caps, i_flushing_caps;     /* mask of dirtied fields */
261de57606cSSage Weil 	struct list_head i_dirty_item, i_flushing_item;
262de57606cSSage Weil 	u64 i_cap_flush_seq;
263de57606cSSage Weil 	/* we need to track cap writeback on a per-cap-bit basis, to allow
264de57606cSSage Weil 	 * overlapping, pipelined cap flushes to the mds.  we can probably
265de57606cSSage Weil 	 * reduce the tid to 8 bits if we're concerned about inode size. */
266de57606cSSage Weil 	u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS];
267de57606cSSage Weil 	wait_queue_head_t i_cap_wq;      /* threads waiting on a capability */
268de57606cSSage Weil 	unsigned long i_hold_caps_min; /* jiffies */
269de57606cSSage Weil 	unsigned long i_hold_caps_max; /* jiffies */
270de57606cSSage Weil 	struct list_head i_cap_delay_list;  /* for delayed cap release to mds */
271de57606cSSage Weil 	int i_cap_exporting_mds;         /* to handle cap migration between */
272de57606cSSage Weil 	unsigned i_cap_exporting_mseq;   /*  mds's. */
273de57606cSSage Weil 	unsigned i_cap_exporting_issued;
274de57606cSSage Weil 	struct ceph_cap_reservation i_cap_migration_resv;
275de57606cSSage Weil 	struct list_head i_cap_snaps;   /* snapped state pending flush to mds */
2767d8cb26dSSage Weil 	struct ceph_snap_context *i_head_snapc;  /* set if wr_buffer_head > 0 or
2777d8cb26dSSage Weil 						    dirty|flushing caps */
278de57606cSSage Weil 	unsigned i_snap_caps;           /* cap bits for snapped files */
279de57606cSSage Weil 
280de57606cSSage Weil 	int i_nr_by_mode[CEPH_FILE_MODE_NUM];  /* open file counts */
281de57606cSSage Weil 
282de57606cSSage Weil 	u32 i_truncate_seq;        /* last truncate to smaller size */
283de57606cSSage Weil 	u64 i_truncate_size;       /*  and the size we last truncated down to */
284de57606cSSage Weil 	int i_truncate_pending;    /*  still need to call vmtruncate */
285de57606cSSage Weil 
286de57606cSSage Weil 	u64 i_max_size;            /* max file size authorized by mds */
287de57606cSSage Weil 	u64 i_reported_size; /* (max_)size reported to or requested of mds */
288de57606cSSage Weil 	u64 i_wanted_max_size;     /* offset we'd like to write too */
289de57606cSSage Weil 	u64 i_requested_max_size;  /* max_size we've requested */
290de57606cSSage Weil 
291de57606cSSage Weil 	/* held references to caps */
292de57606cSSage Weil 	int i_pin_ref;
293de57606cSSage Weil 	int i_rd_ref, i_rdcache_ref, i_wr_ref;
294de57606cSSage Weil 	int i_wrbuffer_ref, i_wrbuffer_ref_head;
295de57606cSSage Weil 	u32 i_shared_gen;       /* increment each time we get FILE_SHARED */
296*cd045cb4SSage Weil 	u32 i_rdcache_gen;      /* incremented each time we get FILE_CACHE. */
297de57606cSSage Weil 	u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
298de57606cSSage Weil 
299de57606cSSage Weil 	struct list_head i_unsafe_writes; /* uncommitted sync writes */
300de57606cSSage Weil 	struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
301de57606cSSage Weil 	spinlock_t i_unsafe_lock;
302de57606cSSage Weil 
303de57606cSSage Weil 	struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
304de57606cSSage Weil 	int i_snap_realm_counter; /* snap realm (if caps) */
305de57606cSSage Weil 	struct list_head i_snap_realm_item;
306de57606cSSage Weil 	struct list_head i_snap_flush_item;
307de57606cSSage Weil 
308de57606cSSage Weil 	struct work_struct i_wb_work;  /* writeback work */
309de57606cSSage Weil 	struct work_struct i_pg_inv_work;  /* page invalidation work */
310de57606cSSage Weil 
311de57606cSSage Weil 	struct work_struct i_vmtruncate_work;
312de57606cSSage Weil 
313de57606cSSage Weil 	struct inode vfs_inode; /* at end */
314de57606cSSage Weil };
315de57606cSSage Weil 
316de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
317de57606cSSage Weil {
318fbbccec9SNoah Watkins 	return container_of(inode, struct ceph_inode_info, vfs_inode);
319de57606cSSage Weil }
320de57606cSSage Weil 
3213d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode)
3223d14c5d2SYehuda Sadeh {
3233d14c5d2SYehuda Sadeh 	return ceph_inode(inode)->i_vino;
3243d14c5d2SYehuda Sadeh }
3253d14c5d2SYehuda Sadeh 
3263d14c5d2SYehuda Sadeh /*
3273d14c5d2SYehuda Sadeh  * ino_t is <64 bits on many architectures, blech.
3283d14c5d2SYehuda Sadeh  *
3293d14c5d2SYehuda Sadeh  * don't include snap in ino hash, at least for now.
3303d14c5d2SYehuda Sadeh  */
3313d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
3323d14c5d2SYehuda Sadeh {
3333d14c5d2SYehuda Sadeh 	ino_t ino = (ino_t)vino.ino;  /* ^ (vino.snap << 20); */
3343d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32
3353d14c5d2SYehuda Sadeh 	ino ^= vino.ino >> (sizeof(u64)-sizeof(ino_t)) * 8;
3363d14c5d2SYehuda Sadeh 	if (!ino)
3373d14c5d2SYehuda Sadeh 		ino = 1;
3383d14c5d2SYehuda Sadeh #endif
3393d14c5d2SYehuda Sadeh 	return ino;
3403d14c5d2SYehuda Sadeh }
3413d14c5d2SYehuda Sadeh 
3423d14c5d2SYehuda Sadeh /* for printf-style formatting */
3433d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
3443d14c5d2SYehuda Sadeh 
3453d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode)
3463d14c5d2SYehuda Sadeh {
3473d14c5d2SYehuda Sadeh 	return ceph_inode(inode)->i_vino.ino;
3483d14c5d2SYehuda Sadeh }
3493d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode)
3503d14c5d2SYehuda Sadeh {
3513d14c5d2SYehuda Sadeh 	return ceph_inode(inode)->i_vino.snap;
3523d14c5d2SYehuda Sadeh }
3533d14c5d2SYehuda Sadeh 
3543d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data)
3553d14c5d2SYehuda Sadeh {
3563d14c5d2SYehuda Sadeh 	struct ceph_vino *pvino = (struct ceph_vino *)data;
3573d14c5d2SYehuda Sadeh 	struct ceph_inode_info *ci = ceph_inode(inode);
3583d14c5d2SYehuda Sadeh 	return ci->i_vino.ino == pvino->ino &&
3593d14c5d2SYehuda Sadeh 		ci->i_vino.snap == pvino->snap;
3603d14c5d2SYehuda Sadeh }
3613d14c5d2SYehuda Sadeh 
3623d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb,
3633d14c5d2SYehuda Sadeh 					    struct ceph_vino vino)
3643d14c5d2SYehuda Sadeh {
3653d14c5d2SYehuda Sadeh 	ino_t t = ceph_vino_to_ino(vino);
3663d14c5d2SYehuda Sadeh 	return ilookup5(sb, t, ceph_ino_compare, &vino);
3673d14c5d2SYehuda Sadeh }
3683d14c5d2SYehuda Sadeh 
3693d14c5d2SYehuda Sadeh 
3703d14c5d2SYehuda Sadeh /*
3713d14c5d2SYehuda Sadeh  * Ceph inode.
3723d14c5d2SYehuda Sadeh  */
3733d14c5d2SYehuda Sadeh #define CEPH_I_COMPLETE  1  /* we have complete directory cached */
3743d14c5d2SYehuda Sadeh #define CEPH_I_NODELAY   4  /* do not delay cap release */
3753d14c5d2SYehuda Sadeh #define CEPH_I_FLUSH     8  /* do not delay flush of dirty metadata */
3763d14c5d2SYehuda Sadeh #define CEPH_I_NOFLUSH  16  /* do not flush dirty caps */
3773d14c5d2SYehuda Sadeh 
378de57606cSSage Weil static inline void ceph_i_clear(struct inode *inode, unsigned mask)
379de57606cSSage Weil {
380de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
381de57606cSSage Weil 
382de57606cSSage Weil 	spin_lock(&inode->i_lock);
383de57606cSSage Weil 	ci->i_ceph_flags &= ~mask;
384de57606cSSage Weil 	spin_unlock(&inode->i_lock);
385de57606cSSage Weil }
386de57606cSSage Weil 
387de57606cSSage Weil static inline void ceph_i_set(struct inode *inode, unsigned mask)
388de57606cSSage Weil {
389de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
390de57606cSSage Weil 
391de57606cSSage Weil 	spin_lock(&inode->i_lock);
392de57606cSSage Weil 	ci->i_ceph_flags |= mask;
393de57606cSSage Weil 	spin_unlock(&inode->i_lock);
394de57606cSSage Weil }
395de57606cSSage Weil 
396de57606cSSage Weil static inline bool ceph_i_test(struct inode *inode, unsigned mask)
397de57606cSSage Weil {
398de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
399de57606cSSage Weil 	bool r;
400de57606cSSage Weil 
401efa4c120SSage Weil 	spin_lock(&inode->i_lock);
402de57606cSSage Weil 	r = (ci->i_ceph_flags & mask) == mask;
403efa4c120SSage Weil 	spin_unlock(&inode->i_lock);
404de57606cSSage Weil 	return r;
405de57606cSSage Weil }
406de57606cSSage Weil 
407de57606cSSage Weil 
408de57606cSSage Weil /* find a specific frag @f */
409de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
410de57606cSSage Weil 						u32 f);
411de57606cSSage Weil 
412de57606cSSage Weil /*
413de57606cSSage Weil  * choose fragment for value @v.  copy frag content to pfrag, if leaf
414de57606cSSage Weil  * exists
415de57606cSSage Weil  */
416de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
417de57606cSSage Weil 			    struct ceph_inode_frag *pfrag,
418de57606cSSage Weil 			    int *found);
419de57606cSSage Weil 
420de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
421de57606cSSage Weil {
422de57606cSSage Weil 	return (struct ceph_dentry_info *)dentry->d_fsdata;
423de57606cSSage Weil }
424de57606cSSage Weil 
425de57606cSSage Weil static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
426de57606cSSage Weil {
427de57606cSSage Weil 	return ((loff_t)frag << 32) | (loff_t)off;
428de57606cSSage Weil }
429de57606cSSage Weil 
430de57606cSSage Weil static inline int ceph_set_ino_cb(struct inode *inode, void *data)
431de57606cSSage Weil {
432de57606cSSage Weil 	ceph_inode(inode)->i_vino = *(struct ceph_vino *)data;
433de57606cSSage Weil 	inode->i_ino = ceph_vino_to_ino(*(struct ceph_vino *)data);
434de57606cSSage Weil 	return 0;
435de57606cSSage Weil }
436de57606cSSage Weil 
437de57606cSSage Weil /*
438de57606cSSage Weil  * caps helpers
439de57606cSSage Weil  */
440de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
441de57606cSSage Weil {
442de57606cSSage Weil 	return !RB_EMPTY_ROOT(&ci->i_caps);
443de57606cSSage Weil }
444de57606cSSage Weil 
445de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
446de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
447de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
448de57606cSSage Weil 				    struct ceph_cap *cap);
449de57606cSSage Weil 
450de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci)
451de57606cSSage Weil {
452de57606cSSage Weil 	int issued;
453de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
454de57606cSSage Weil 	issued = __ceph_caps_issued(ci, NULL);
455de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
456de57606cSSage Weil 	return issued;
457de57606cSSage Weil }
458de57606cSSage Weil 
459de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
460de57606cSSage Weil 					int touch)
461de57606cSSage Weil {
462de57606cSSage Weil 	int r;
463de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
464de57606cSSage Weil 	r = __ceph_caps_issued_mask(ci, mask, touch);
465de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
466de57606cSSage Weil 	return r;
467de57606cSSage Weil }
468de57606cSSage Weil 
469de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
470de57606cSSage Weil {
471de57606cSSage Weil 	return ci->i_dirty_caps | ci->i_flushing_caps;
472de57606cSSage Weil }
473afcdaea3SSage Weil extern void __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask);
474de57606cSSage Weil 
475de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
476de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci);
477de57606cSSage Weil 
478de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
479de57606cSSage Weil 
480de57606cSSage Weil /*
481de57606cSSage Weil  * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
482de57606cSSage Weil  */
483de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
484de57606cSSage Weil {
485de57606cSSage Weil 	int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
486de57606cSSage Weil 	if (w & CEPH_CAP_FILE_BUFFER)
487de57606cSSage Weil 		w |= CEPH_CAP_FILE_EXCL;  /* we want EXCL if dirty data */
488de57606cSSage Weil 	return w;
489de57606cSSage Weil }
490de57606cSSage Weil 
491de57606cSSage Weil /* what the mds thinks we want */
492de57606cSSage Weil extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
493de57606cSSage Weil 
49437151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc);
49537151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
49637151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
49737151668SYehuda Sadeh extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
49837151668SYehuda Sadeh 			     struct ceph_cap_reservation *ctx, int need);
49937151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
50037151668SYehuda Sadeh 			       struct ceph_cap_reservation *ctx);
5013d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client,
502de57606cSSage Weil 				    int *total, int *avail, int *used,
50385ccce43SSage Weil 				    int *reserved, int *min);
504de57606cSSage Weil 
5053d14c5d2SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode)
506de57606cSSage Weil {
5073d14c5d2SYehuda Sadeh 	return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
508de57606cSSage Weil }
509de57606cSSage Weil 
5103d14c5d2SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb)
511de57606cSSage Weil {
5123d14c5d2SYehuda Sadeh 	return (struct ceph_fs_client *)sb->s_fs_info;
513de57606cSSage Weil }
514de57606cSSage Weil 
515de57606cSSage Weil 
516de57606cSSage Weil /*
517de57606cSSage Weil  * we keep buffered readdir results attached to file->private_data
518de57606cSSage Weil  */
519de57606cSSage Weil struct ceph_file_info {
520de57606cSSage Weil 	int fmode;     /* initialized on open */
521de57606cSSage Weil 
522de57606cSSage Weil 	/* readdir: position within the dir */
523de57606cSSage Weil 	u32 frag;
524de57606cSSage Weil 	struct ceph_mds_request *last_readdir;
525de57606cSSage Weil 	int at_end;
526de57606cSSage Weil 
527de57606cSSage Weil 	/* readdir: position within a frag */
528de57606cSSage Weil 	unsigned offset;       /* offset of last chunk, adjusted for . and .. */
529de57606cSSage Weil 	u64 next_offset;       /* offset of next chunk (last_name's + 1) */
530de57606cSSage Weil 	char *last_name;       /* last entry in previous chunk */
531de57606cSSage Weil 	struct dentry *dentry; /* next dentry (for dcache readdir) */
532de57606cSSage Weil 	unsigned long dir_release_count;
533de57606cSSage Weil 
534de57606cSSage Weil 	/* used for -o dirstat read() on directory thing */
535de57606cSSage Weil 	char *dir_info;
536de57606cSSage Weil 	int dir_info_len;
537de57606cSSage Weil };
538de57606cSSage Weil 
539de57606cSSage Weil 
540de57606cSSage Weil 
541de57606cSSage Weil /*
542de57606cSSage Weil  * A "snap realm" describes a subset of the file hierarchy sharing
543de57606cSSage Weil  * the same set of snapshots that apply to it.  The realms themselves
544de57606cSSage Weil  * are organized into a hierarchy, such that children inherit (some of)
545de57606cSSage Weil  * the snapshots of their parents.
546de57606cSSage Weil  *
547de57606cSSage Weil  * All inodes within the realm that have capabilities are linked into a
548de57606cSSage Weil  * per-realm list.
549de57606cSSage Weil  */
550de57606cSSage Weil struct ceph_snap_realm {
551de57606cSSage Weil 	u64 ino;
552de57606cSSage Weil 	atomic_t nref;
553a105f00cSSage Weil 	struct rb_node node;
554a105f00cSSage Weil 
555de57606cSSage Weil 	u64 created, seq;
556de57606cSSage Weil 	u64 parent_ino;
557de57606cSSage Weil 	u64 parent_since;   /* snapid when our current parent became so */
558de57606cSSage Weil 
559de57606cSSage Weil 	u64 *prior_parent_snaps;      /* snaps inherited from any parents we */
560de57606cSSage Weil 	int num_prior_parent_snaps;   /*  had prior to parent_since */
561de57606cSSage Weil 	u64 *snaps;                   /* snaps specific to this realm */
562de57606cSSage Weil 	int num_snaps;
563de57606cSSage Weil 
564de57606cSSage Weil 	struct ceph_snap_realm *parent;
565de57606cSSage Weil 	struct list_head children;       /* list of child realms */
566de57606cSSage Weil 	struct list_head child_item;
567de57606cSSage Weil 
568de57606cSSage Weil 	struct list_head empty_item;     /* if i have ref==0 */
569de57606cSSage Weil 
570ae00d4f3SSage Weil 	struct list_head dirty_item;     /* if realm needs new context */
571ae00d4f3SSage Weil 
572de57606cSSage Weil 	/* the current set of snaps for this realm */
573de57606cSSage Weil 	struct ceph_snap_context *cached_context;
574de57606cSSage Weil 
575de57606cSSage Weil 	struct list_head inodes_with_caps;
576de57606cSSage Weil 	spinlock_t inodes_with_caps_lock;
577de57606cSSage Weil };
578de57606cSSage Weil 
5793d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void)
5803d14c5d2SYehuda Sadeh {
5813d14c5d2SYehuda Sadeh 	int congestion_kb;
582de57606cSSage Weil 
583de57606cSSage Weil 	/*
5843d14c5d2SYehuda Sadeh 	 * Copied from NFS
5853d14c5d2SYehuda Sadeh 	 *
5863d14c5d2SYehuda Sadeh 	 * congestion size, scale with available memory.
5873d14c5d2SYehuda Sadeh 	 *
5883d14c5d2SYehuda Sadeh 	 *  64MB:    8192k
5893d14c5d2SYehuda Sadeh 	 * 128MB:   11585k
5903d14c5d2SYehuda Sadeh 	 * 256MB:   16384k
5913d14c5d2SYehuda Sadeh 	 * 512MB:   23170k
5923d14c5d2SYehuda Sadeh 	 *   1GB:   32768k
5933d14c5d2SYehuda Sadeh 	 *   2GB:   46340k
5943d14c5d2SYehuda Sadeh 	 *   4GB:   65536k
5953d14c5d2SYehuda Sadeh 	 *   8GB:   92681k
5963d14c5d2SYehuda Sadeh 	 *  16GB:  131072k
5973d14c5d2SYehuda Sadeh 	 *
5983d14c5d2SYehuda Sadeh 	 * This allows larger machines to have larger/more transfers.
5993d14c5d2SYehuda Sadeh 	 * Limit the default to 256M
600de57606cSSage Weil 	 */
6013d14c5d2SYehuda Sadeh 	congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
6023d14c5d2SYehuda Sadeh 	if (congestion_kb > 256*1024)
6033d14c5d2SYehuda Sadeh 		congestion_kb = 256*1024;
6043d14c5d2SYehuda Sadeh 
6053d14c5d2SYehuda Sadeh 	return congestion_kb;
606de57606cSSage Weil }
607de57606cSSage Weil 
608de57606cSSage Weil 
609de57606cSSage Weil 
610de57606cSSage Weil /* snap.c */
611de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
612de57606cSSage Weil 					       u64 ino);
613de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
614de57606cSSage Weil 				struct ceph_snap_realm *realm);
615de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
616de57606cSSage Weil 				struct ceph_snap_realm *realm);
617de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m,
618de57606cSSage Weil 				  void *p, void *e, bool deletion);
619de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
6202600d2ddSSage Weil 			     struct ceph_mds_session *session,
621de57606cSSage Weil 			     struct ceph_msg *msg);
622fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
623de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
624de57606cSSage Weil 				  struct ceph_cap_snap *capsnap);
625de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
626de57606cSSage Weil 
627de57606cSSage Weil /*
628de57606cSSage Weil  * a cap_snap is "pending" if it is still awaiting an in-progress
629de57606cSSage Weil  * sync write (that may/may not still update size, mtime, etc.).
630de57606cSSage Weil  */
631de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
632de57606cSSage Weil {
633de57606cSSage Weil 	return !list_empty(&ci->i_cap_snaps) &&
634de57606cSSage Weil 		list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap,
635de57606cSSage Weil 			   ci_item)->writing;
636de57606cSSage Weil }
637de57606cSSage Weil 
638de57606cSSage Weil /* inode.c */
639de57606cSSage Weil extern const struct inode_operations ceph_file_iops;
640de57606cSSage Weil 
641de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb);
642de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode);
643de57606cSSage Weil 
644de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb,
645de57606cSSage Weil 				    struct ceph_vino vino);
646de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent);
647de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued,
648de57606cSSage Weil 			       u32 truncate_seq, u64 truncate_size, u64 size);
649de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued,
650de57606cSSage Weil 				u64 time_warp_seq, struct timespec *ctime,
651de57606cSSage Weil 				struct timespec *mtime, struct timespec *atime);
652de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb,
653de57606cSSage Weil 			   struct ceph_mds_request *req,
654de57606cSSage Weil 			   struct ceph_mds_session *session);
655de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
656de57606cSSage Weil 				    struct ceph_mds_session *session);
657de57606cSSage Weil 
658de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask);
659de57606cSSage Weil 
660de57606cSSage Weil extern int ceph_inode_set_size(struct inode *inode, loff_t size);
661de57606cSSage Weil extern void __ceph_do_pending_vmtruncate(struct inode *inode);
6623c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode);
6633c6f6b79SSage Weil 
6643c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode);
6653c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode);
666de57606cSSage Weil 
667de57606cSSage Weil extern int ceph_do_getattr(struct inode *inode, int mask);
668de57606cSSage Weil extern int ceph_permission(struct inode *inode, int mask);
669de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
670de57606cSSage Weil extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
671de57606cSSage Weil 			struct kstat *stat);
672de57606cSSage Weil 
673de57606cSSage Weil /* xattr.c */
674de57606cSSage Weil extern int ceph_setxattr(struct dentry *, const char *, const void *,
675de57606cSSage Weil 			 size_t, int);
676de57606cSSage Weil extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t);
677de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
678de57606cSSage Weil extern int ceph_removexattr(struct dentry *, const char *);
679de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
680de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
681de57606cSSage Weil 
682de57606cSSage Weil /* caps.c */
683de57606cSSage Weil extern const char *ceph_cap_string(int c);
684de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session,
685de57606cSSage Weil 			     struct ceph_msg *msg);
686de57606cSSage Weil extern int ceph_add_cap(struct inode *inode,
687de57606cSSage Weil 			struct ceph_mds_session *session, u64 cap_id,
688de57606cSSage Weil 			int fmode, unsigned issued, unsigned wanted,
689de57606cSSage Weil 			unsigned cap, unsigned seq, u64 realmino, int flags,
690de57606cSSage Weil 			struct ceph_cap_reservation *caps_reservation);
6917c1332b8SSage Weil extern void __ceph_remove_cap(struct ceph_cap *cap);
692de57606cSSage Weil static inline void ceph_remove_cap(struct ceph_cap *cap)
693de57606cSSage Weil {
694de57606cSSage Weil 	struct inode *inode = &cap->ci->vfs_inode;
695de57606cSSage Weil 	spin_lock(&inode->i_lock);
6967c1332b8SSage Weil 	__ceph_remove_cap(cap);
697de57606cSSage Weil 	spin_unlock(&inode->i_lock);
698de57606cSSage Weil }
69937151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc,
70037151668SYehuda Sadeh 			 struct ceph_cap *cap);
701de57606cSSage Weil 
702de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode);
703f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
7047ea80859SChristoph Hellwig extern int ceph_fsync(struct file *file, int datasync);
705de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
706de57606cSSage Weil 				    struct ceph_mds_session *session);
7072bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
7082bc50259SGreg Farnum 					     int mds);
709de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode);
710de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
711de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
712de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
713de57606cSSage Weil 				       struct ceph_snap_context *snapc);
714de57606cSSage Weil extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
715e835124cSSage Weil 			       struct ceph_mds_session **psession,
716e835124cSSage Weil 			       int again);
717de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
718de57606cSSage Weil 			    struct ceph_mds_session *session);
719afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
720afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
721de57606cSSage Weil 
722de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode,
723de57606cSSage Weil 				     int mds, int drop, int unless, int force);
724de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
725de57606cSSage Weil 				      int mds, int drop, int unless);
726de57606cSSage Weil 
727de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
728de57606cSSage Weil 			 int *got, loff_t endoff);
729de57606cSSage Weil 
730de57606cSSage Weil /* for counting open files by mode */
731de57606cSSage Weil static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
732de57606cSSage Weil {
733de57606cSSage Weil 	ci->i_nr_by_mode[mode]++;
734de57606cSSage Weil }
735de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
736de57606cSSage Weil 
737de57606cSSage Weil /* addr.c */
738de57606cSSage Weil extern const struct address_space_operations ceph_aops;
739de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
740de57606cSSage Weil 
741de57606cSSage Weil /* file.c */
742de57606cSSage Weil extern const struct file_operations ceph_file_fops;
743de57606cSSage Weil extern const struct address_space_operations ceph_aops;
7443d14c5d2SYehuda Sadeh extern int ceph_copy_to_page_vector(struct page **pages,
7453d14c5d2SYehuda Sadeh 				    const char *data,
7463d14c5d2SYehuda Sadeh 				    loff_t off, size_t len);
7473d14c5d2SYehuda Sadeh extern int ceph_copy_from_page_vector(struct page **pages,
7483d14c5d2SYehuda Sadeh 				    char *data,
7493d14c5d2SYehuda Sadeh 				    loff_t off, size_t len);
7503d14c5d2SYehuda Sadeh extern struct page **ceph_alloc_page_vector(int num_pages, gfp_t flags);
751de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file);
752de57606cSSage Weil extern struct dentry *ceph_lookup_open(struct inode *dir, struct dentry *dentry,
753de57606cSSage Weil 				       struct nameidata *nd, int mode,
754de57606cSSage Weil 				       int locked_dir);
755de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp);
756de57606cSSage Weil 
757de57606cSSage Weil /* dir.c */
758de57606cSSage Weil extern const struct file_operations ceph_dir_fops;
759de57606cSSage Weil extern const struct inode_operations ceph_dir_iops;
76052dfb8acSSage Weil extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
761de57606cSSage Weil 	ceph_snapdir_dentry_ops;
762de57606cSSage Weil 
763de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
764de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
765de57606cSSage Weil 					 struct dentry *dentry, int err);
766de57606cSSage Weil 
767de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn);
768de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn);
769de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn);
77081a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
771de57606cSSage Weil 
772de57606cSSage Weil /*
773de57606cSSage Weil  * our d_ops vary depending on whether the inode is live,
774de57606cSSage Weil  * snapshotted (read-only), or a virtual ".snap" directory.
775de57606cSSage Weil  */
776de57606cSSage Weil int ceph_init_dentry(struct dentry *dentry);
777de57606cSSage Weil 
778de57606cSSage Weil 
779de57606cSSage Weil /* ioctl.c */
780de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
781de57606cSSage Weil 
782de57606cSSage Weil /* export.c */
783de57606cSSage Weil extern const struct export_operations ceph_export_ops;
784de57606cSSage Weil 
78540819f6fSGreg Farnum /* locks.c */
78640819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
78740819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
78840819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
78940819f6fSGreg Farnum extern int ceph_encode_locks(struct inode *i, struct ceph_pagelist *p,
79040819f6fSGreg Farnum 			     int p_locks, int f_locks);
79140819f6fSGreg Farnum extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c);
79240819f6fSGreg Farnum 
793de57606cSSage Weil static inline struct inode *get_dentry_parent_inode(struct dentry *dentry)
794de57606cSSage Weil {
795de57606cSSage Weil 	if (dentry && dentry->d_parent)
796de57606cSSage Weil 		return dentry->d_parent->d_inode;
797de57606cSSage Weil 
798de57606cSSage Weil 	return NULL;
799de57606cSSage Weil }
800de57606cSSage Weil 
8013d14c5d2SYehuda Sadeh /* debugfs.c */
8023d14c5d2SYehuda Sadeh extern int ceph_fs_debugfs_init(struct ceph_fs_client *client);
8033d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
8043d14c5d2SYehuda Sadeh 
805de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */
806