xref: /linux/fs/ceph/super.h (revision e835124c2be289515b918f2688ced4249e2de566)
1de57606cSSage Weil #ifndef _FS_CEPH_SUPER_H
2de57606cSSage Weil #define _FS_CEPH_SUPER_H
3de57606cSSage Weil 
4de57606cSSage Weil #include "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 
17de57606cSSage Weil #include "types.h"
18de57606cSSage Weil #include "messenger.h"
19de57606cSSage Weil #include "msgpool.h"
20de57606cSSage Weil #include "mon_client.h"
21de57606cSSage Weil #include "mds_client.h"
22de57606cSSage Weil #include "osd_client.h"
23de57606cSSage Weil #include "ceph_fs.h"
24de57606cSSage Weil 
25de57606cSSage Weil /* f_type in struct statfs */
26de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400
27de57606cSSage Weil 
28de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate
29de57606cSSage Weil  * large volume sizes on 32-bit machines. */
30de57606cSSage Weil #define CEPH_BLOCK_SHIFT   20  /* 1 MB */
31de57606cSSage Weil #define CEPH_BLOCK         (1 << CEPH_BLOCK_SHIFT)
32de57606cSSage Weil 
33de57606cSSage Weil /*
346a259382SSage Weil  * Supported features
356a259382SSage Weil  */
36c6f3fdc5SGreg Farnum #define CEPH_FEATURE_SUPPORTED CEPH_FEATURE_NOSRCADDR | CEPH_FEATURE_FLOCK
376a259382SSage Weil #define CEPH_FEATURE_REQUIRED  CEPH_FEATURE_NOSRCADDR
386a259382SSage Weil 
396a259382SSage Weil /*
40de57606cSSage Weil  * mount options
41de57606cSSage Weil  */
42de57606cSSage Weil #define CEPH_OPT_FSID             (1<<0)
43de57606cSSage Weil #define CEPH_OPT_NOSHARE          (1<<1) /* don't share client with other sbs */
44de57606cSSage Weil #define CEPH_OPT_MYIP             (1<<2) /* specified my ip */
45de57606cSSage Weil #define CEPH_OPT_DIRSTAT          (1<<4) /* funky `cat dirname` for stats */
46de57606cSSage Weil #define CEPH_OPT_RBYTES           (1<<5) /* dir st_bytes = rbytes */
47de57606cSSage Weil #define CEPH_OPT_NOCRC            (1<<6) /* no data crc on writes */
48de57606cSSage Weil #define CEPH_OPT_NOASYNCREADDIR   (1<<7) /* no dcache readdir */
49de57606cSSage Weil 
50de57606cSSage Weil #define CEPH_OPT_DEFAULT   (CEPH_OPT_RBYTES)
51de57606cSSage Weil 
52de57606cSSage Weil #define ceph_set_opt(client, opt) \
536b805185SSage Weil 	(client)->mount_args->flags |= CEPH_OPT_##opt;
54de57606cSSage Weil #define ceph_test_opt(client, opt) \
556b805185SSage Weil 	(!!((client)->mount_args->flags & CEPH_OPT_##opt))
56de57606cSSage Weil 
57de57606cSSage Weil 
58de57606cSSage Weil struct ceph_mount_args {
59de57606cSSage Weil 	int sb_flags;
60de57606cSSage Weil 	int flags;
61de57606cSSage Weil 	struct ceph_fsid fsid;
62de57606cSSage Weil 	struct ceph_entity_addr my_addr;
636e19a16eSSage Weil 	int num_mon;
646e19a16eSSage Weil 	struct ceph_entity_addr *mon_addr;
656e19a16eSSage Weil 	int mount_timeout;
666e19a16eSSage Weil 	int osd_idle_ttl;
67de57606cSSage Weil 	int osd_timeout;
68422d2cb8SYehuda Sadeh 	int osd_keepalive_timeout;
696e19a16eSSage Weil 	int wsize;
706e19a16eSSage Weil 	int rsize;            /* max readahead */
716e19a16eSSage Weil 	int congestion_kb;    /* max writeback in flight */
726e19a16eSSage Weil 	int caps_wanted_delay_min, caps_wanted_delay_max;
736e19a16eSSage Weil 	int cap_release_safety;
7423804d91SSage Weil 	int max_readdir;       /* max readdir result (entires) */
7523804d91SSage Weil 	int max_readdir_bytes; /* max readdir result (bytes) */
76de57606cSSage Weil 	char *snapdir_name;   /* default ".snap" */
774e7a5dcdSSage Weil 	char *name;
78de57606cSSage Weil 	char *secret;
79de57606cSSage Weil };
80de57606cSSage Weil 
81de57606cSSage Weil /*
82de57606cSSage Weil  * defaults
83de57606cSSage Weil  */
84de57606cSSage Weil #define CEPH_MOUNT_TIMEOUT_DEFAULT  60
85422d2cb8SYehuda Sadeh #define CEPH_OSD_TIMEOUT_DEFAULT    60  /* seconds */
86422d2cb8SYehuda Sadeh #define CEPH_OSD_KEEPALIVE_DEFAULT  5
87f5a2041bSYehuda Sadeh #define CEPH_OSD_IDLE_TTL_DEFAULT    60
888fa97655SSage Weil #define CEPH_MOUNT_RSIZE_DEFAULT    (512*1024) /* readahead */
896e19a16eSSage Weil #define CEPH_MAX_READDIR_DEFAULT    1024
9023804d91SSage Weil #define CEPH_MAX_READDIR_BYTES_DEFAULT    (512*1024)
91de57606cSSage Weil 
92de57606cSSage Weil #define CEPH_MSG_MAX_FRONT_LEN	(16*1024*1024)
93de57606cSSage Weil #define CEPH_MSG_MAX_DATA_LEN	(16*1024*1024)
94de57606cSSage Weil 
95de57606cSSage Weil #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
964e7a5dcdSSage Weil #define CEPH_AUTH_NAME_DEFAULT   "guest"
97de57606cSSage Weil /*
98de57606cSSage Weil  * Delay telling the MDS we no longer want caps, in case we reopen
99de57606cSSage Weil  * the file.  Delay a minimum amount of time, even if we send a cap
100de57606cSSage Weil  * message for some other reason.  Otherwise, take the oppotunity to
101de57606cSSage Weil  * update the mds to avoid sending another message later.
102de57606cSSage Weil  */
103de57606cSSage Weil #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT      5  /* cap release delay */
104de57606cSSage Weil #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT     60  /* cap release delay */
105de57606cSSage Weil 
1066e19a16eSSage Weil #define CEPH_CAP_RELEASE_SAFETY_DEFAULT        (CEPH_CAPS_PER_RELEASE * 4)
107de57606cSSage Weil 
108de57606cSSage Weil /* mount state */
109de57606cSSage Weil enum {
110de57606cSSage Weil 	CEPH_MOUNT_MOUNTING,
111de57606cSSage Weil 	CEPH_MOUNT_MOUNTED,
112de57606cSSage Weil 	CEPH_MOUNT_UNMOUNTING,
113de57606cSSage Weil 	CEPH_MOUNT_UNMOUNTED,
114de57606cSSage Weil 	CEPH_MOUNT_SHUTDOWN,
115de57606cSSage Weil };
116de57606cSSage Weil 
117de57606cSSage Weil /*
118de57606cSSage Weil  * subtract jiffies
119de57606cSSage Weil  */
120de57606cSSage Weil static inline unsigned long time_sub(unsigned long a, unsigned long b)
121de57606cSSage Weil {
122de57606cSSage Weil 	BUG_ON(time_after(b, a));
123de57606cSSage Weil 	return (long)a - (long)b;
124de57606cSSage Weil }
125de57606cSSage Weil 
126de57606cSSage Weil /*
127de57606cSSage Weil  * per-filesystem client state
128de57606cSSage Weil  *
129de57606cSSage Weil  * possibly shared by multiple mount points, if they are
130de57606cSSage Weil  * mounting the same ceph filesystem/cluster.
131de57606cSSage Weil  */
132de57606cSSage Weil struct ceph_client {
1330743304dSSage Weil 	struct ceph_fsid fsid;
1340743304dSSage Weil 	bool have_fsid;
135de57606cSSage Weil 
136de57606cSSage Weil 	struct mutex mount_mutex;       /* serialize mount attempts */
1376b805185SSage Weil 	struct ceph_mount_args *mount_args;
138de57606cSSage Weil 
139de57606cSSage Weil 	struct super_block *sb;
140de57606cSSage Weil 
141de57606cSSage Weil 	unsigned long mount_state;
1429bd2e6f8SSage Weil 	wait_queue_head_t auth_wq;
143de57606cSSage Weil 
1449bd2e6f8SSage Weil 	int auth_err;
145de57606cSSage Weil 
14685ccce43SSage Weil 	int min_caps;                  /* min caps i added */
14785ccce43SSage Weil 
148de57606cSSage Weil 	struct ceph_messenger *msgr;   /* messenger instance */
149de57606cSSage Weil 	struct ceph_mon_client monc;
150de57606cSSage Weil 	struct ceph_mds_client mdsc;
151de57606cSSage Weil 	struct ceph_osd_client osdc;
152de57606cSSage Weil 
153de57606cSSage Weil 	/* writeback */
154de57606cSSage Weil 	mempool_t *wb_pagevec_pool;
155de57606cSSage Weil 	struct workqueue_struct *wb_wq;
156de57606cSSage Weil 	struct workqueue_struct *pg_inv_wq;
157de57606cSSage Weil 	struct workqueue_struct *trunc_wq;
1582baba250SYehuda Sadeh 	atomic_long_t writeback_count;
159de57606cSSage Weil 
160de57606cSSage Weil 	struct backing_dev_info backing_dev_info;
1610743304dSSage Weil 
1620743304dSSage Weil #ifdef CONFIG_DEBUG_FS
1630743304dSSage Weil 	struct dentry *debugfs_monmap;
1640743304dSSage Weil 	struct dentry *debugfs_mdsmap, *debugfs_osdmap;
1650743304dSSage Weil 	struct dentry *debugfs_dir, *debugfs_dentry_lru, *debugfs_caps;
1662baba250SYehuda Sadeh 	struct dentry *debugfs_congestion_kb;
16706edf046SSage Weil 	struct dentry *debugfs_bdi;
1680743304dSSage Weil #endif
169de57606cSSage Weil };
170de57606cSSage Weil 
171de57606cSSage Weil /*
172de57606cSSage Weil  * File i/o capability.  This tracks shared state with the metadata
173de57606cSSage Weil  * server that allows us to cache or writeback attributes or to read
174de57606cSSage Weil  * and write data.  For any given inode, we should have one or more
175de57606cSSage Weil  * capabilities, one issued by each metadata server, and our
176de57606cSSage Weil  * cumulative access is the OR of all issued capabilities.
177de57606cSSage Weil  *
178de57606cSSage Weil  * Each cap is referenced by the inode's i_caps rbtree and by per-mds
179de57606cSSage Weil  * session capability lists.
180de57606cSSage Weil  */
181de57606cSSage Weil struct ceph_cap {
182de57606cSSage Weil 	struct ceph_inode_info *ci;
183de57606cSSage Weil 	struct rb_node ci_node;          /* per-ci cap tree */
184de57606cSSage Weil 	struct ceph_mds_session *session;
185de57606cSSage Weil 	struct list_head session_caps;   /* per-session caplist */
186de57606cSSage Weil 	int mds;
187de57606cSSage Weil 	u64 cap_id;       /* unique cap id (mds provided) */
188de57606cSSage Weil 	int issued;       /* latest, from the mds */
189de57606cSSage Weil 	int implemented;  /* implemented superset of issued (for revocation) */
190de57606cSSage Weil 	int mds_wanted;
191685f9a5dSSage Weil 	u32 seq, issue_seq, mseq;
192685f9a5dSSage Weil 	u32 cap_gen;      /* active/stale cycle */
193de57606cSSage Weil 	unsigned long last_used;
194de57606cSSage Weil 	struct list_head caps_item;
195de57606cSSage Weil };
196de57606cSSage Weil 
197de57606cSSage Weil #define CHECK_CAPS_NODELAY    1  /* do not delay any further */
198de57606cSSage Weil #define CHECK_CAPS_AUTHONLY   2  /* only check auth cap */
199de57606cSSage Weil #define CHECK_CAPS_FLUSH      4  /* flush any dirty caps */
200de57606cSSage Weil 
201de57606cSSage Weil /*
202de57606cSSage Weil  * Snapped cap state that is pending flush to mds.  When a snapshot occurs,
203de57606cSSage Weil  * we first complete any in-process sync writes and writeback any dirty
204de57606cSSage Weil  * data before flushing the snapped state (tracked here) back to the MDS.
205de57606cSSage Weil  */
206de57606cSSage Weil struct ceph_cap_snap {
207de57606cSSage Weil 	atomic_t nref;
208de57606cSSage Weil 	struct ceph_inode_info *ci;
209de57606cSSage Weil 	struct list_head ci_item, flushing_item;
210de57606cSSage Weil 
211de57606cSSage Weil 	u64 follows, flush_tid;
212de57606cSSage Weil 	int issued, dirty;
213de57606cSSage Weil 	struct ceph_snap_context *context;
214de57606cSSage Weil 
215de57606cSSage Weil 	mode_t mode;
216de57606cSSage Weil 	uid_t uid;
217de57606cSSage Weil 	gid_t gid;
218de57606cSSage Weil 
2194a625be4SSage Weil 	struct ceph_buffer *xattr_blob;
220de57606cSSage Weil 	u64 xattr_version;
221de57606cSSage Weil 
222de57606cSSage Weil 	u64 size;
223de57606cSSage Weil 	struct timespec mtime, atime, ctime;
224de57606cSSage Weil 	u64 time_warp_seq;
225de57606cSSage Weil 	int writing;   /* a sync write is still in progress */
226de57606cSSage Weil 	int dirty_pages;     /* dirty pages awaiting writeback */
227de57606cSSage Weil };
228de57606cSSage Weil 
229de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
230de57606cSSage Weil {
2314a625be4SSage Weil 	if (atomic_dec_and_test(&capsnap->nref)) {
2324a625be4SSage Weil 		if (capsnap->xattr_blob)
2334a625be4SSage Weil 			ceph_buffer_put(capsnap->xattr_blob);
234de57606cSSage Weil 		kfree(capsnap);
235de57606cSSage Weil 	}
2364a625be4SSage Weil }
237de57606cSSage Weil 
238de57606cSSage Weil /*
239de57606cSSage Weil  * The frag tree describes how a directory is fragmented, potentially across
240de57606cSSage Weil  * multiple metadata servers.  It is also used to indicate points where
241de57606cSSage Weil  * metadata authority is delegated, and whether/where metadata is replicated.
242de57606cSSage Weil  *
243de57606cSSage Weil  * A _leaf_ frag will be present in the i_fragtree IFF there is
244de57606cSSage Weil  * delegation info.  That is, if mds >= 0 || ndist > 0.
245de57606cSSage Weil  */
246de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4
247de57606cSSage Weil 
248de57606cSSage Weil struct ceph_inode_frag {
249de57606cSSage Weil 	struct rb_node node;
250de57606cSSage Weil 
251de57606cSSage Weil 	/* fragtree state */
252de57606cSSage Weil 	u32 frag;
253de57606cSSage Weil 	int split_by;         /* i.e. 2^(split_by) children */
254de57606cSSage Weil 
255de57606cSSage Weil 	/* delegation and replication info */
256de57606cSSage Weil 	int mds;              /* -1 if same authority as parent */
257de57606cSSage Weil 	int ndist;            /* >0 if replicated */
258de57606cSSage Weil 	int dist[CEPH_MAX_DIRFRAG_REP];
259de57606cSSage Weil };
260de57606cSSage Weil 
261de57606cSSage Weil /*
262de57606cSSage Weil  * We cache inode xattrs as an encoded blob until they are first used,
263de57606cSSage Weil  * at which point we parse them into an rbtree.
264de57606cSSage Weil  */
265de57606cSSage Weil struct ceph_inode_xattr {
266de57606cSSage Weil 	struct rb_node node;
267de57606cSSage Weil 
268de57606cSSage Weil 	const char *name;
269de57606cSSage Weil 	int name_len;
270de57606cSSage Weil 	const char *val;
271de57606cSSage Weil 	int val_len;
272de57606cSSage Weil 	int dirty;
273de57606cSSage Weil 
274de57606cSSage Weil 	int should_free_name;
275de57606cSSage Weil 	int should_free_val;
276de57606cSSage Weil };
277de57606cSSage Weil 
278de57606cSSage Weil struct ceph_inode_xattrs_info {
279de57606cSSage Weil 	/*
280de57606cSSage Weil 	 * (still encoded) xattr blob. we avoid the overhead of parsing
281de57606cSSage Weil 	 * this until someone actually calls getxattr, etc.
282de57606cSSage Weil 	 *
283de57606cSSage Weil 	 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
284de57606cSSage Weil 	 * NULL means we don't know.
285de57606cSSage Weil 	*/
286de57606cSSage Weil 	struct ceph_buffer *blob, *prealloc_blob;
287de57606cSSage Weil 
288de57606cSSage Weil 	struct rb_root index;
289de57606cSSage Weil 	bool dirty;
290de57606cSSage Weil 	int count;
291de57606cSSage Weil 	int names_size;
292de57606cSSage Weil 	int vals_size;
293de57606cSSage Weil 	u64 version, index_version;
294de57606cSSage Weil };
295de57606cSSage Weil 
296de57606cSSage Weil /*
297de57606cSSage Weil  * Ceph inode.
298de57606cSSage Weil  */
299de57606cSSage Weil #define CEPH_I_COMPLETE  1  /* we have complete directory cached */
300de57606cSSage Weil #define CEPH_I_NODELAY   4  /* do not delay cap release */
301de57606cSSage Weil #define CEPH_I_FLUSH     8  /* do not delay flush of dirty metadata */
302e9964c10SSage Weil #define CEPH_I_NOFLUSH  16  /* do not flush dirty caps */
303de57606cSSage Weil 
304de57606cSSage Weil struct ceph_inode_info {
305de57606cSSage Weil 	struct ceph_vino i_vino;   /* ceph ino + snap */
306de57606cSSage Weil 
307de57606cSSage Weil 	u64 i_version;
308de57606cSSage Weil 	u32 i_time_warp_seq;
309de57606cSSage Weil 
310de57606cSSage Weil 	unsigned i_ceph_flags;
311de57606cSSage Weil 	unsigned long i_release_count;
312de57606cSSage Weil 
313de57606cSSage Weil 	struct ceph_file_layout i_layout;
314de57606cSSage Weil 	char *i_symlink;
315de57606cSSage Weil 
316de57606cSSage Weil 	/* for dirs */
317de57606cSSage Weil 	struct timespec i_rctime;
318de57606cSSage Weil 	u64 i_rbytes, i_rfiles, i_rsubdirs;
319de57606cSSage Weil 	u64 i_files, i_subdirs;
320de57606cSSage Weil 	u64 i_max_offset;  /* largest readdir offset, set with I_COMPLETE */
321de57606cSSage Weil 
322de57606cSSage Weil 	struct rb_root i_fragtree;
323de57606cSSage Weil 	struct mutex i_fragtree_mutex;
324de57606cSSage Weil 
325de57606cSSage Weil 	struct ceph_inode_xattrs_info i_xattrs;
326de57606cSSage Weil 
327de57606cSSage Weil 	/* capabilities.  protected _both_ by i_lock and cap->session's
328de57606cSSage Weil 	 * s_mutex. */
329de57606cSSage Weil 	struct rb_root i_caps;           /* cap list */
330de57606cSSage Weil 	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
331de57606cSSage Weil 	unsigned i_dirty_caps, i_flushing_caps;     /* mask of dirtied fields */
332de57606cSSage Weil 	struct list_head i_dirty_item, i_flushing_item;
333de57606cSSage Weil 	u64 i_cap_flush_seq;
334de57606cSSage Weil 	/* we need to track cap writeback on a per-cap-bit basis, to allow
335de57606cSSage Weil 	 * overlapping, pipelined cap flushes to the mds.  we can probably
336de57606cSSage Weil 	 * reduce the tid to 8 bits if we're concerned about inode size. */
337de57606cSSage Weil 	u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS];
338de57606cSSage Weil 	wait_queue_head_t i_cap_wq;      /* threads waiting on a capability */
339de57606cSSage Weil 	unsigned long i_hold_caps_min; /* jiffies */
340de57606cSSage Weil 	unsigned long i_hold_caps_max; /* jiffies */
341de57606cSSage Weil 	struct list_head i_cap_delay_list;  /* for delayed cap release to mds */
342de57606cSSage Weil 	int i_cap_exporting_mds;         /* to handle cap migration between */
343de57606cSSage Weil 	unsigned i_cap_exporting_mseq;   /*  mds's. */
344de57606cSSage Weil 	unsigned i_cap_exporting_issued;
345de57606cSSage Weil 	struct ceph_cap_reservation i_cap_migration_resv;
346de57606cSSage Weil 	struct list_head i_cap_snaps;   /* snapped state pending flush to mds */
3477d8cb26dSSage Weil 	struct ceph_snap_context *i_head_snapc;  /* set if wr_buffer_head > 0 or
3487d8cb26dSSage Weil 						    dirty|flushing caps */
349de57606cSSage Weil 	unsigned i_snap_caps;           /* cap bits for snapped files */
350de57606cSSage Weil 
351de57606cSSage Weil 	int i_nr_by_mode[CEPH_FILE_MODE_NUM];  /* open file counts */
352de57606cSSage Weil 
353de57606cSSage Weil 	u32 i_truncate_seq;        /* last truncate to smaller size */
354de57606cSSage Weil 	u64 i_truncate_size;       /*  and the size we last truncated down to */
355de57606cSSage Weil 	int i_truncate_pending;    /*  still need to call vmtruncate */
356de57606cSSage Weil 
357de57606cSSage Weil 	u64 i_max_size;            /* max file size authorized by mds */
358de57606cSSage Weil 	u64 i_reported_size; /* (max_)size reported to or requested of mds */
359de57606cSSage Weil 	u64 i_wanted_max_size;     /* offset we'd like to write too */
360de57606cSSage Weil 	u64 i_requested_max_size;  /* max_size we've requested */
361de57606cSSage Weil 
362de57606cSSage Weil 	/* held references to caps */
363de57606cSSage Weil 	int i_pin_ref;
364de57606cSSage Weil 	int i_rd_ref, i_rdcache_ref, i_wr_ref;
365de57606cSSage Weil 	int i_wrbuffer_ref, i_wrbuffer_ref_head;
366de57606cSSage Weil 	u32 i_shared_gen;       /* increment each time we get FILE_SHARED */
367de57606cSSage Weil 	u32 i_rdcache_gen;      /* we increment this each time we get
368de57606cSSage Weil 				   FILE_CACHE.  If it's non-zero, we
369de57606cSSage Weil 				   _may_ have cached pages. */
370de57606cSSage Weil 	u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
371de57606cSSage Weil 
372de57606cSSage Weil 	struct list_head i_unsafe_writes; /* uncommitted sync writes */
373de57606cSSage Weil 	struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
374de57606cSSage Weil 	spinlock_t i_unsafe_lock;
375de57606cSSage Weil 
376de57606cSSage Weil 	struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
377de57606cSSage Weil 	int i_snap_realm_counter; /* snap realm (if caps) */
378de57606cSSage Weil 	struct list_head i_snap_realm_item;
379de57606cSSage Weil 	struct list_head i_snap_flush_item;
380de57606cSSage Weil 
381de57606cSSage Weil 	struct work_struct i_wb_work;  /* writeback work */
382de57606cSSage Weil 	struct work_struct i_pg_inv_work;  /* page invalidation work */
383de57606cSSage Weil 
384de57606cSSage Weil 	struct work_struct i_vmtruncate_work;
385de57606cSSage Weil 
386de57606cSSage Weil 	struct inode vfs_inode; /* at end */
387de57606cSSage Weil };
388de57606cSSage Weil 
389de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
390de57606cSSage Weil {
391fbbccec9SNoah Watkins 	return container_of(inode, struct ceph_inode_info, vfs_inode);
392de57606cSSage Weil }
393de57606cSSage Weil 
394de57606cSSage Weil static inline void ceph_i_clear(struct inode *inode, unsigned mask)
395de57606cSSage Weil {
396de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
397de57606cSSage Weil 
398de57606cSSage Weil 	spin_lock(&inode->i_lock);
399de57606cSSage Weil 	ci->i_ceph_flags &= ~mask;
400de57606cSSage Weil 	spin_unlock(&inode->i_lock);
401de57606cSSage Weil }
402de57606cSSage Weil 
403de57606cSSage Weil static inline void ceph_i_set(struct inode *inode, unsigned mask)
404de57606cSSage Weil {
405de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
406de57606cSSage Weil 
407de57606cSSage Weil 	spin_lock(&inode->i_lock);
408de57606cSSage Weil 	ci->i_ceph_flags |= mask;
409de57606cSSage Weil 	spin_unlock(&inode->i_lock);
410de57606cSSage Weil }
411de57606cSSage Weil 
412de57606cSSage Weil static inline bool ceph_i_test(struct inode *inode, unsigned mask)
413de57606cSSage Weil {
414de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
415de57606cSSage Weil 	bool r;
416de57606cSSage Weil 
417de57606cSSage Weil 	smp_mb();
418de57606cSSage Weil 	r = (ci->i_ceph_flags & mask) == mask;
419de57606cSSage Weil 	return r;
420de57606cSSage Weil }
421de57606cSSage Weil 
422de57606cSSage Weil 
423de57606cSSage Weil /* find a specific frag @f */
424de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
425de57606cSSage Weil 						u32 f);
426de57606cSSage Weil 
427de57606cSSage Weil /*
428de57606cSSage Weil  * choose fragment for value @v.  copy frag content to pfrag, if leaf
429de57606cSSage Weil  * exists
430de57606cSSage Weil  */
431de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
432de57606cSSage Weil 			    struct ceph_inode_frag *pfrag,
433de57606cSSage Weil 			    int *found);
434de57606cSSage Weil 
435de57606cSSage Weil /*
436de57606cSSage Weil  * Ceph dentry state
437de57606cSSage Weil  */
438de57606cSSage Weil struct ceph_dentry_info {
439de57606cSSage Weil 	struct ceph_mds_session *lease_session;
440de57606cSSage Weil 	u32 lease_gen, lease_shared_gen;
441de57606cSSage Weil 	u32 lease_seq;
442de57606cSSage Weil 	unsigned long lease_renew_after, lease_renew_from;
443de57606cSSage Weil 	struct list_head lru;
444de57606cSSage Weil 	struct dentry *dentry;
445de57606cSSage Weil 	u64 time;
446de57606cSSage Weil 	u64 offset;
447de57606cSSage Weil };
448de57606cSSage Weil 
449de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
450de57606cSSage Weil {
451de57606cSSage Weil 	return (struct ceph_dentry_info *)dentry->d_fsdata;
452de57606cSSage Weil }
453de57606cSSage Weil 
454de57606cSSage Weil static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
455de57606cSSage Weil {
456de57606cSSage Weil 	return ((loff_t)frag << 32) | (loff_t)off;
457de57606cSSage Weil }
458de57606cSSage Weil 
459de57606cSSage Weil /*
460de57606cSSage Weil  * ino_t is <64 bits on many architectures, blech.
461de57606cSSage Weil  *
462de57606cSSage Weil  * don't include snap in ino hash, at least for now.
463de57606cSSage Weil  */
464de57606cSSage Weil static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
465de57606cSSage Weil {
466de57606cSSage Weil 	ino_t ino = (ino_t)vino.ino;  /* ^ (vino.snap << 20); */
467de57606cSSage Weil #if BITS_PER_LONG == 32
468de57606cSSage Weil 	ino ^= vino.ino >> (sizeof(u64)-sizeof(ino_t)) * 8;
469de57606cSSage Weil 	if (!ino)
470de57606cSSage Weil 		ino = 1;
471de57606cSSage Weil #endif
472de57606cSSage Weil 	return ino;
473de57606cSSage Weil }
474de57606cSSage Weil 
475de57606cSSage Weil static inline int ceph_set_ino_cb(struct inode *inode, void *data)
476de57606cSSage Weil {
477de57606cSSage Weil 	ceph_inode(inode)->i_vino = *(struct ceph_vino *)data;
478de57606cSSage Weil 	inode->i_ino = ceph_vino_to_ino(*(struct ceph_vino *)data);
479de57606cSSage Weil 	return 0;
480de57606cSSage Weil }
481de57606cSSage Weil 
482de57606cSSage Weil static inline struct ceph_vino ceph_vino(struct inode *inode)
483de57606cSSage Weil {
484de57606cSSage Weil 	return ceph_inode(inode)->i_vino;
485de57606cSSage Weil }
486de57606cSSage Weil 
487de57606cSSage Weil /* for printf-style formatting */
488de57606cSSage Weil #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
489de57606cSSage Weil 
490de57606cSSage Weil static inline u64 ceph_ino(struct inode *inode)
491de57606cSSage Weil {
492de57606cSSage Weil 	return ceph_inode(inode)->i_vino.ino;
493de57606cSSage Weil }
494de57606cSSage Weil static inline u64 ceph_snap(struct inode *inode)
495de57606cSSage Weil {
496de57606cSSage Weil 	return ceph_inode(inode)->i_vino.snap;
497de57606cSSage Weil }
498de57606cSSage Weil 
499de57606cSSage Weil static inline int ceph_ino_compare(struct inode *inode, void *data)
500de57606cSSage Weil {
501de57606cSSage Weil 	struct ceph_vino *pvino = (struct ceph_vino *)data;
502de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
503de57606cSSage Weil 	return ci->i_vino.ino == pvino->ino &&
504de57606cSSage Weil 		ci->i_vino.snap == pvino->snap;
505de57606cSSage Weil }
506de57606cSSage Weil 
507de57606cSSage Weil static inline struct inode *ceph_find_inode(struct super_block *sb,
508de57606cSSage Weil 					    struct ceph_vino vino)
509de57606cSSage Weil {
510de57606cSSage Weil 	ino_t t = ceph_vino_to_ino(vino);
511de57606cSSage Weil 	return ilookup5(sb, t, ceph_ino_compare, &vino);
512de57606cSSage Weil }
513de57606cSSage Weil 
514de57606cSSage Weil 
515de57606cSSage Weil /*
516de57606cSSage Weil  * caps helpers
517de57606cSSage Weil  */
518de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
519de57606cSSage Weil {
520de57606cSSage Weil 	return !RB_EMPTY_ROOT(&ci->i_caps);
521de57606cSSage Weil }
522de57606cSSage Weil 
523de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
524de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
525de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
526de57606cSSage Weil 				    struct ceph_cap *cap);
527de57606cSSage Weil 
528de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci)
529de57606cSSage Weil {
530de57606cSSage Weil 	int issued;
531de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
532de57606cSSage Weil 	issued = __ceph_caps_issued(ci, NULL);
533de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
534de57606cSSage Weil 	return issued;
535de57606cSSage Weil }
536de57606cSSage Weil 
537de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
538de57606cSSage Weil 					int touch)
539de57606cSSage Weil {
540de57606cSSage Weil 	int r;
541de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
542de57606cSSage Weil 	r = __ceph_caps_issued_mask(ci, mask, touch);
543de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
544de57606cSSage Weil 	return r;
545de57606cSSage Weil }
546de57606cSSage Weil 
547de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
548de57606cSSage Weil {
549de57606cSSage Weil 	return ci->i_dirty_caps | ci->i_flushing_caps;
550de57606cSSage Weil }
551afcdaea3SSage Weil extern void __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask);
552de57606cSSage Weil 
553de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
554de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci);
555de57606cSSage Weil 
556de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
557de57606cSSage Weil 
558de57606cSSage Weil /*
559de57606cSSage Weil  * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
560de57606cSSage Weil  */
561de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
562de57606cSSage Weil {
563de57606cSSage Weil 	int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
564de57606cSSage Weil 	if (w & CEPH_CAP_FILE_BUFFER)
565de57606cSSage Weil 		w |= CEPH_CAP_FILE_EXCL;  /* we want EXCL if dirty data */
566de57606cSSage Weil 	return w;
567de57606cSSage Weil }
568de57606cSSage Weil 
569de57606cSSage Weil /* what the mds thinks we want */
570de57606cSSage Weil extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
571de57606cSSage Weil 
57237151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc);
57337151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
57437151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
57537151668SYehuda Sadeh extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
57637151668SYehuda Sadeh 			     struct ceph_cap_reservation *ctx, int need);
57737151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
57837151668SYehuda Sadeh 			       struct ceph_cap_reservation *ctx);
579de57606cSSage Weil extern void ceph_reservation_status(struct ceph_client *client,
580de57606cSSage Weil 				    int *total, int *avail, int *used,
58185ccce43SSage Weil 				    int *reserved, int *min);
582de57606cSSage Weil 
583de57606cSSage Weil static inline struct ceph_client *ceph_inode_to_client(struct inode *inode)
584de57606cSSage Weil {
585de57606cSSage Weil 	return (struct ceph_client *)inode->i_sb->s_fs_info;
586de57606cSSage Weil }
587de57606cSSage Weil 
588de57606cSSage Weil static inline struct ceph_client *ceph_sb_to_client(struct super_block *sb)
589de57606cSSage Weil {
590de57606cSSage Weil 	return (struct ceph_client *)sb->s_fs_info;
591de57606cSSage Weil }
592de57606cSSage Weil 
593de57606cSSage Weil 
594de57606cSSage Weil /*
595de57606cSSage Weil  * we keep buffered readdir results attached to file->private_data
596de57606cSSage Weil  */
597de57606cSSage Weil struct ceph_file_info {
598de57606cSSage Weil 	int fmode;     /* initialized on open */
599de57606cSSage Weil 
600de57606cSSage Weil 	/* readdir: position within the dir */
601de57606cSSage Weil 	u32 frag;
602de57606cSSage Weil 	struct ceph_mds_request *last_readdir;
603de57606cSSage Weil 	int at_end;
604de57606cSSage Weil 
605de57606cSSage Weil 	/* readdir: position within a frag */
606de57606cSSage Weil 	unsigned offset;       /* offset of last chunk, adjusted for . and .. */
607de57606cSSage Weil 	u64 next_offset;       /* offset of next chunk (last_name's + 1) */
608de57606cSSage Weil 	char *last_name;       /* last entry in previous chunk */
609de57606cSSage Weil 	struct dentry *dentry; /* next dentry (for dcache readdir) */
610de57606cSSage Weil 	unsigned long dir_release_count;
611de57606cSSage Weil 
612de57606cSSage Weil 	/* used for -o dirstat read() on directory thing */
613de57606cSSage Weil 	char *dir_info;
614de57606cSSage Weil 	int dir_info_len;
615de57606cSSage Weil };
616de57606cSSage Weil 
617de57606cSSage Weil 
618de57606cSSage Weil 
619de57606cSSage Weil /*
620de57606cSSage Weil  * snapshots
621de57606cSSage Weil  */
622de57606cSSage Weil 
623de57606cSSage Weil /*
624de57606cSSage Weil  * A "snap context" is the set of existing snapshots when we
625de57606cSSage Weil  * write data.  It is used by the OSD to guide its COW behavior.
626de57606cSSage Weil  *
627de57606cSSage Weil  * The ceph_snap_context is refcounted, and attached to each dirty
628de57606cSSage Weil  * page, indicating which context the dirty data belonged when it was
629de57606cSSage Weil  * dirtied.
630de57606cSSage Weil  */
631de57606cSSage Weil struct ceph_snap_context {
632de57606cSSage Weil 	atomic_t nref;
633de57606cSSage Weil 	u64 seq;
634de57606cSSage Weil 	int num_snaps;
635de57606cSSage Weil 	u64 snaps[];
636de57606cSSage Weil };
637de57606cSSage Weil 
638de57606cSSage Weil static inline struct ceph_snap_context *
639de57606cSSage Weil ceph_get_snap_context(struct ceph_snap_context *sc)
640de57606cSSage Weil {
641de57606cSSage Weil 	/*
642de57606cSSage Weil 	printk("get_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
643de57606cSSage Weil 	       atomic_read(&sc->nref)+1);
644de57606cSSage Weil 	*/
645de57606cSSage Weil 	if (sc)
646de57606cSSage Weil 		atomic_inc(&sc->nref);
647de57606cSSage Weil 	return sc;
648de57606cSSage Weil }
649de57606cSSage Weil 
650de57606cSSage Weil static inline void ceph_put_snap_context(struct ceph_snap_context *sc)
651de57606cSSage Weil {
652de57606cSSage Weil 	if (!sc)
653de57606cSSage Weil 		return;
654de57606cSSage Weil 	/*
655de57606cSSage Weil 	printk("put_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
656de57606cSSage Weil 	       atomic_read(&sc->nref)-1);
657de57606cSSage Weil 	*/
658de57606cSSage Weil 	if (atomic_dec_and_test(&sc->nref)) {
659de57606cSSage Weil 		/*printk(" deleting snap_context %p\n", sc);*/
660de57606cSSage Weil 		kfree(sc);
661de57606cSSage Weil 	}
662de57606cSSage Weil }
663de57606cSSage Weil 
664de57606cSSage Weil /*
665de57606cSSage Weil  * A "snap realm" describes a subset of the file hierarchy sharing
666de57606cSSage Weil  * the same set of snapshots that apply to it.  The realms themselves
667de57606cSSage Weil  * are organized into a hierarchy, such that children inherit (some of)
668de57606cSSage Weil  * the snapshots of their parents.
669de57606cSSage Weil  *
670de57606cSSage Weil  * All inodes within the realm that have capabilities are linked into a
671de57606cSSage Weil  * per-realm list.
672de57606cSSage Weil  */
673de57606cSSage Weil struct ceph_snap_realm {
674de57606cSSage Weil 	u64 ino;
675de57606cSSage Weil 	atomic_t nref;
676a105f00cSSage Weil 	struct rb_node node;
677a105f00cSSage Weil 
678de57606cSSage Weil 	u64 created, seq;
679de57606cSSage Weil 	u64 parent_ino;
680de57606cSSage Weil 	u64 parent_since;   /* snapid when our current parent became so */
681de57606cSSage Weil 
682de57606cSSage Weil 	u64 *prior_parent_snaps;      /* snaps inherited from any parents we */
683de57606cSSage Weil 	int num_prior_parent_snaps;   /*  had prior to parent_since */
684de57606cSSage Weil 	u64 *snaps;                   /* snaps specific to this realm */
685de57606cSSage Weil 	int num_snaps;
686de57606cSSage Weil 
687de57606cSSage Weil 	struct ceph_snap_realm *parent;
688de57606cSSage Weil 	struct list_head children;       /* list of child realms */
689de57606cSSage Weil 	struct list_head child_item;
690de57606cSSage Weil 
691de57606cSSage Weil 	struct list_head empty_item;     /* if i have ref==0 */
692de57606cSSage Weil 
693ae00d4f3SSage Weil 	struct list_head dirty_item;     /* if realm needs new context */
694ae00d4f3SSage Weil 
695de57606cSSage Weil 	/* the current set of snaps for this realm */
696de57606cSSage Weil 	struct ceph_snap_context *cached_context;
697de57606cSSage Weil 
698de57606cSSage Weil 	struct list_head inodes_with_caps;
699de57606cSSage Weil 	spinlock_t inodes_with_caps_lock;
700de57606cSSage Weil };
701de57606cSSage Weil 
702de57606cSSage Weil 
703de57606cSSage Weil 
704de57606cSSage Weil /*
705de57606cSSage Weil  * calculate the number of pages a given length and offset map onto,
706de57606cSSage Weil  * if we align the data.
707de57606cSSage Weil  */
708de57606cSSage Weil static inline int calc_pages_for(u64 off, u64 len)
709de57606cSSage Weil {
710de57606cSSage Weil 	return ((off+len+PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT) -
711de57606cSSage Weil 		(off >> PAGE_CACHE_SHIFT);
712de57606cSSage Weil }
713de57606cSSage Weil 
714de57606cSSage Weil 
715de57606cSSage Weil 
716de57606cSSage Weil /* snap.c */
717de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
718de57606cSSage Weil 					       u64 ino);
719de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
720de57606cSSage Weil 				struct ceph_snap_realm *realm);
721de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
722de57606cSSage Weil 				struct ceph_snap_realm *realm);
723de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m,
724de57606cSSage Weil 				  void *p, void *e, bool deletion);
725de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
7262600d2ddSSage Weil 			     struct ceph_mds_session *session,
727de57606cSSage Weil 			     struct ceph_msg *msg);
728fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
729de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
730de57606cSSage Weil 				  struct ceph_cap_snap *capsnap);
731de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
732de57606cSSage Weil 
733de57606cSSage Weil /*
734de57606cSSage Weil  * a cap_snap is "pending" if it is still awaiting an in-progress
735de57606cSSage Weil  * sync write (that may/may not still update size, mtime, etc.).
736de57606cSSage Weil  */
737de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
738de57606cSSage Weil {
739de57606cSSage Weil 	return !list_empty(&ci->i_cap_snaps) &&
740de57606cSSage Weil 		list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap,
741de57606cSSage Weil 			   ci_item)->writing;
742de57606cSSage Weil }
743de57606cSSage Weil 
744de57606cSSage Weil 
745de57606cSSage Weil /* super.c */
746de57606cSSage Weil extern struct kmem_cache *ceph_inode_cachep;
747de57606cSSage Weil extern struct kmem_cache *ceph_cap_cachep;
748de57606cSSage Weil extern struct kmem_cache *ceph_dentry_cachep;
749de57606cSSage Weil extern struct kmem_cache *ceph_file_cachep;
750de57606cSSage Weil 
751de57606cSSage Weil extern const char *ceph_msg_type_name(int type);
7520743304dSSage Weil extern int ceph_check_fsid(struct ceph_client *client, struct ceph_fsid *fsid);
753de57606cSSage Weil 
754de57606cSSage Weil /* inode.c */
755de57606cSSage Weil extern const struct inode_operations ceph_file_iops;
756de57606cSSage Weil 
757de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb);
758de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode);
759de57606cSSage Weil 
760de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb,
761de57606cSSage Weil 				    struct ceph_vino vino);
762de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent);
763de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued,
764de57606cSSage Weil 			       u32 truncate_seq, u64 truncate_size, u64 size);
765de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued,
766de57606cSSage Weil 				u64 time_warp_seq, struct timespec *ctime,
767de57606cSSage Weil 				struct timespec *mtime, struct timespec *atime);
768de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb,
769de57606cSSage Weil 			   struct ceph_mds_request *req,
770de57606cSSage Weil 			   struct ceph_mds_session *session);
771de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
772de57606cSSage Weil 				    struct ceph_mds_session *session);
773de57606cSSage Weil 
774de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask);
775de57606cSSage Weil 
776de57606cSSage Weil extern int ceph_inode_set_size(struct inode *inode, loff_t size);
777de57606cSSage Weil extern void __ceph_do_pending_vmtruncate(struct inode *inode);
7783c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode);
7793c6f6b79SSage Weil 
7803c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode);
7813c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode);
782de57606cSSage Weil 
783de57606cSSage Weil extern int ceph_do_getattr(struct inode *inode, int mask);
784de57606cSSage Weil extern int ceph_permission(struct inode *inode, int mask);
785de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
786de57606cSSage Weil extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
787de57606cSSage Weil 			struct kstat *stat);
788de57606cSSage Weil 
789de57606cSSage Weil /* xattr.c */
790de57606cSSage Weil extern int ceph_setxattr(struct dentry *, const char *, const void *,
791de57606cSSage Weil 			 size_t, int);
792de57606cSSage Weil extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t);
793de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
794de57606cSSage Weil extern int ceph_removexattr(struct dentry *, const char *);
795de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
796de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
797de57606cSSage Weil 
798de57606cSSage Weil /* caps.c */
799de57606cSSage Weil extern const char *ceph_cap_string(int c);
800de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session,
801de57606cSSage Weil 			     struct ceph_msg *msg);
802de57606cSSage Weil extern int ceph_add_cap(struct inode *inode,
803de57606cSSage Weil 			struct ceph_mds_session *session, u64 cap_id,
804de57606cSSage Weil 			int fmode, unsigned issued, unsigned wanted,
805de57606cSSage Weil 			unsigned cap, unsigned seq, u64 realmino, int flags,
806de57606cSSage Weil 			struct ceph_cap_reservation *caps_reservation);
8077c1332b8SSage Weil extern void __ceph_remove_cap(struct ceph_cap *cap);
808de57606cSSage Weil static inline void ceph_remove_cap(struct ceph_cap *cap)
809de57606cSSage Weil {
810de57606cSSage Weil 	struct inode *inode = &cap->ci->vfs_inode;
811de57606cSSage Weil 	spin_lock(&inode->i_lock);
8127c1332b8SSage Weil 	__ceph_remove_cap(cap);
813de57606cSSage Weil 	spin_unlock(&inode->i_lock);
814de57606cSSage Weil }
81537151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc,
81637151668SYehuda Sadeh 			 struct ceph_cap *cap);
817de57606cSSage Weil 
818de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode);
819f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
8207ea80859SChristoph Hellwig extern int ceph_fsync(struct file *file, int datasync);
821de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
822de57606cSSage Weil 				    struct ceph_mds_session *session);
8232bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
8242bc50259SGreg Farnum 					     int mds);
825de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode);
826de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
827de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
828de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
829de57606cSSage Weil 				       struct ceph_snap_context *snapc);
830de57606cSSage Weil extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
831*e835124cSSage Weil 			       struct ceph_mds_session **psession,
832*e835124cSSage Weil 			       int again);
833de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
834de57606cSSage Weil 			    struct ceph_mds_session *session);
835afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
836afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
837de57606cSSage Weil 
838de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode,
839de57606cSSage Weil 				     int mds, int drop, int unless, int force);
840de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
841de57606cSSage Weil 				      int mds, int drop, int unless);
842de57606cSSage Weil 
843de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
844de57606cSSage Weil 			 int *got, loff_t endoff);
845de57606cSSage Weil 
846de57606cSSage Weil /* for counting open files by mode */
847de57606cSSage Weil static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
848de57606cSSage Weil {
849de57606cSSage Weil 	ci->i_nr_by_mode[mode]++;
850de57606cSSage Weil }
851de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
852de57606cSSage Weil 
853de57606cSSage Weil /* addr.c */
854de57606cSSage Weil extern const struct address_space_operations ceph_aops;
855de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
856de57606cSSage Weil 
857de57606cSSage Weil /* file.c */
858de57606cSSage Weil extern const struct file_operations ceph_file_fops;
859de57606cSSage Weil extern const struct address_space_operations ceph_aops;
860de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file);
861de57606cSSage Weil extern struct dentry *ceph_lookup_open(struct inode *dir, struct dentry *dentry,
862de57606cSSage Weil 				       struct nameidata *nd, int mode,
863de57606cSSage Weil 				       int locked_dir);
864de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp);
865de57606cSSage Weil extern void ceph_release_page_vector(struct page **pages, int num_pages);
866de57606cSSage Weil 
867de57606cSSage Weil /* dir.c */
868de57606cSSage Weil extern const struct file_operations ceph_dir_fops;
869de57606cSSage Weil extern const struct inode_operations ceph_dir_iops;
87052dfb8acSSage Weil extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
871de57606cSSage Weil 	ceph_snapdir_dentry_ops;
872de57606cSSage Weil 
873de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
874de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
875de57606cSSage Weil 					 struct dentry *dentry, int err);
876de57606cSSage Weil 
877de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn);
878de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn);
879de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn);
88081a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
881de57606cSSage Weil 
882de57606cSSage Weil /*
883de57606cSSage Weil  * our d_ops vary depending on whether the inode is live,
884de57606cSSage Weil  * snapshotted (read-only), or a virtual ".snap" directory.
885de57606cSSage Weil  */
886de57606cSSage Weil int ceph_init_dentry(struct dentry *dentry);
887de57606cSSage Weil 
888de57606cSSage Weil 
889de57606cSSage Weil /* ioctl.c */
890de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
891de57606cSSage Weil 
892de57606cSSage Weil /* export.c */
893de57606cSSage Weil extern const struct export_operations ceph_export_ops;
894de57606cSSage Weil 
895de57606cSSage Weil /* debugfs.c */
896de57606cSSage Weil extern int ceph_debugfs_init(void);
897de57606cSSage Weil extern void ceph_debugfs_cleanup(void);
898de57606cSSage Weil extern int ceph_debugfs_client_init(struct ceph_client *client);
899de57606cSSage Weil extern void ceph_debugfs_client_cleanup(struct ceph_client *client);
900de57606cSSage Weil 
90140819f6fSGreg Farnum /* locks.c */
90240819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
90340819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
90440819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
90540819f6fSGreg Farnum extern int ceph_encode_locks(struct inode *i, struct ceph_pagelist *p,
90640819f6fSGreg Farnum 			     int p_locks, int f_locks);
90740819f6fSGreg Farnum extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c);
90840819f6fSGreg Farnum 
909de57606cSSage Weil static inline struct inode *get_dentry_parent_inode(struct dentry *dentry)
910de57606cSSage Weil {
911de57606cSSage Weil 	if (dentry && dentry->d_parent)
912de57606cSSage Weil 		return dentry->d_parent->d_inode;
913de57606cSSage Weil 
914de57606cSSage Weil 	return NULL;
915de57606cSSage Weil }
916de57606cSSage Weil 
917de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */
918