xref: /linux/fs/ceph/super.h (revision 8fa9765576875200a7412a5300b5f0537211f038)
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>
14de57606cSSage Weil 
15de57606cSSage Weil #include "types.h"
16de57606cSSage Weil #include "messenger.h"
17de57606cSSage Weil #include "msgpool.h"
18de57606cSSage Weil #include "mon_client.h"
19de57606cSSage Weil #include "mds_client.h"
20de57606cSSage Weil #include "osd_client.h"
21de57606cSSage Weil #include "ceph_fs.h"
22de57606cSSage Weil 
23de57606cSSage Weil /* f_type in struct statfs */
24de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400
25de57606cSSage Weil 
26de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate
27de57606cSSage Weil  * large volume sizes on 32-bit machines. */
28de57606cSSage Weil #define CEPH_BLOCK_SHIFT   20  /* 1 MB */
29de57606cSSage Weil #define CEPH_BLOCK         (1 << CEPH_BLOCK_SHIFT)
30de57606cSSage Weil 
31de57606cSSage Weil /*
32de57606cSSage Weil  * mount options
33de57606cSSage Weil  */
34de57606cSSage Weil #define CEPH_OPT_FSID             (1<<0)
35de57606cSSage Weil #define CEPH_OPT_NOSHARE          (1<<1) /* don't share client with other sbs */
36de57606cSSage Weil #define CEPH_OPT_MYIP             (1<<2) /* specified my ip */
37de57606cSSage Weil #define CEPH_OPT_DIRSTAT          (1<<4) /* funky `cat dirname` for stats */
38de57606cSSage Weil #define CEPH_OPT_RBYTES           (1<<5) /* dir st_bytes = rbytes */
39de57606cSSage Weil #define CEPH_OPT_NOCRC            (1<<6) /* no data crc on writes */
40de57606cSSage Weil #define CEPH_OPT_NOASYNCREADDIR   (1<<7) /* no dcache readdir */
41de57606cSSage Weil 
42de57606cSSage Weil #define CEPH_OPT_DEFAULT   (CEPH_OPT_RBYTES)
43de57606cSSage Weil 
44de57606cSSage Weil #define ceph_set_opt(client, opt) \
45de57606cSSage Weil 	(client)->mount_args.flags |= CEPH_OPT_##opt;
46de57606cSSage Weil #define ceph_test_opt(client, opt) \
47de57606cSSage Weil 	(!!((client)->mount_args.flags & CEPH_OPT_##opt))
48de57606cSSage Weil 
49de57606cSSage Weil 
50de57606cSSage Weil #define CEPH_MAX_MON_MOUNT_ADDR	5
51de57606cSSage Weil 
52de57606cSSage Weil struct ceph_mount_args {
53de57606cSSage Weil 	int sb_flags;
54de57606cSSage Weil 	int flags;
55de57606cSSage Weil 	int mount_timeout;
56de57606cSSage Weil 	int caps_wanted_delay_min, caps_wanted_delay_max;
57de57606cSSage Weil 	struct ceph_fsid fsid;
58de57606cSSage Weil 	struct ceph_entity_addr my_addr;
59de57606cSSage Weil 	int wsize;
60de57606cSSage Weil 	int rsize;            /* max readahead */
61de57606cSSage Weil 	int max_readdir;      /* max readdir size */
62de57606cSSage Weil 	int osd_timeout;
63de57606cSSage Weil 	char *snapdir_name;   /* default ".snap" */
64de57606cSSage Weil 	char *secret;
65de57606cSSage Weil 	int cap_release_safety;
66de57606cSSage Weil };
67de57606cSSage Weil 
68de57606cSSage Weil /*
69de57606cSSage Weil  * defaults
70de57606cSSage Weil  */
71de57606cSSage Weil #define CEPH_MOUNT_TIMEOUT_DEFAULT  60
72*8fa97655SSage Weil #define CEPH_MOUNT_RSIZE_DEFAULT    (512*1024) /* readahead */
73de57606cSSage Weil 
74de57606cSSage Weil #define CEPH_MSG_MAX_FRONT_LEN	(16*1024*1024)
75de57606cSSage Weil #define CEPH_MSG_MAX_DATA_LEN	(16*1024*1024)
76de57606cSSage Weil 
77de57606cSSage Weil #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
78de57606cSSage Weil 
79de57606cSSage Weil /*
80de57606cSSage Weil  * Delay telling the MDS we no longer want caps, in case we reopen
81de57606cSSage Weil  * the file.  Delay a minimum amount of time, even if we send a cap
82de57606cSSage Weil  * message for some other reason.  Otherwise, take the oppotunity to
83de57606cSSage Weil  * update the mds to avoid sending another message later.
84de57606cSSage Weil  */
85de57606cSSage Weil #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT      5  /* cap release delay */
86de57606cSSage Weil #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT     60  /* cap release delay */
87de57606cSSage Weil 
88de57606cSSage Weil 
89de57606cSSage Weil /* mount state */
90de57606cSSage Weil enum {
91de57606cSSage Weil 	CEPH_MOUNT_MOUNTING,
92de57606cSSage Weil 	CEPH_MOUNT_MOUNTED,
93de57606cSSage Weil 	CEPH_MOUNT_UNMOUNTING,
94de57606cSSage Weil 	CEPH_MOUNT_UNMOUNTED,
95de57606cSSage Weil 	CEPH_MOUNT_SHUTDOWN,
96de57606cSSage Weil };
97de57606cSSage Weil 
98de57606cSSage Weil /*
99de57606cSSage Weil  * subtract jiffies
100de57606cSSage Weil  */
101de57606cSSage Weil static inline unsigned long time_sub(unsigned long a, unsigned long b)
102de57606cSSage Weil {
103de57606cSSage Weil 	BUG_ON(time_after(b, a));
104de57606cSSage Weil 	return (long)a - (long)b;
105de57606cSSage Weil }
106de57606cSSage Weil 
107de57606cSSage Weil /*
108de57606cSSage Weil  * per-filesystem client state
109de57606cSSage Weil  *
110de57606cSSage Weil  * possibly shared by multiple mount points, if they are
111de57606cSSage Weil  * mounting the same ceph filesystem/cluster.
112de57606cSSage Weil  */
113de57606cSSage Weil struct ceph_client {
114de57606cSSage Weil 	__s64 whoami;                   /* my client number */
115de57606cSSage Weil 	struct dentry *debugfs_monmap;
116de57606cSSage Weil 	struct dentry *debugfs_mdsmap, *debugfs_osdmap;
117de57606cSSage Weil 	struct dentry *debugfs_dir, *debugfs_dentry_lru, *debugfs_caps;
118de57606cSSage Weil 
119de57606cSSage Weil 	struct mutex mount_mutex;       /* serialize mount attempts */
120de57606cSSage Weil 	struct ceph_mount_args mount_args;
121de57606cSSage Weil 	struct ceph_fsid fsid;
122de57606cSSage Weil 
123de57606cSSage Weil 	struct super_block *sb;
124de57606cSSage Weil 
125de57606cSSage Weil 	unsigned long mount_state;
126de57606cSSage Weil 	wait_queue_head_t mount_wq;
127de57606cSSage Weil 
128de57606cSSage Weil 	int mount_err;
129de57606cSSage Weil 	void *signed_ticket;           /* our keys to the kingdom */
130de57606cSSage Weil 	int signed_ticket_len;
131de57606cSSage Weil 
132de57606cSSage Weil 	struct ceph_messenger *msgr;   /* messenger instance */
133de57606cSSage Weil 	struct ceph_mon_client monc;
134de57606cSSage Weil 	struct ceph_mds_client mdsc;
135de57606cSSage Weil 	struct ceph_osd_client osdc;
136de57606cSSage Weil 
137de57606cSSage Weil 	/* writeback */
138de57606cSSage Weil 	mempool_t *wb_pagevec_pool;
139de57606cSSage Weil 	struct workqueue_struct *wb_wq;
140de57606cSSage Weil 	struct workqueue_struct *pg_inv_wq;
141de57606cSSage Weil 	struct workqueue_struct *trunc_wq;
142de57606cSSage Weil 
143de57606cSSage Weil 	struct backing_dev_info backing_dev_info;
144de57606cSSage Weil };
145de57606cSSage Weil 
146de57606cSSage Weil static inline struct ceph_client *ceph_client(struct super_block *sb)
147de57606cSSage Weil {
148de57606cSSage Weil 	return sb->s_fs_info;
149de57606cSSage Weil }
150de57606cSSage Weil 
151de57606cSSage Weil 
152de57606cSSage Weil /*
153de57606cSSage Weil  * File i/o capability.  This tracks shared state with the metadata
154de57606cSSage Weil  * server that allows us to cache or writeback attributes or to read
155de57606cSSage Weil  * and write data.  For any given inode, we should have one or more
156de57606cSSage Weil  * capabilities, one issued by each metadata server, and our
157de57606cSSage Weil  * cumulative access is the OR of all issued capabilities.
158de57606cSSage Weil  *
159de57606cSSage Weil  * Each cap is referenced by the inode's i_caps rbtree and by per-mds
160de57606cSSage Weil  * session capability lists.
161de57606cSSage Weil  */
162de57606cSSage Weil struct ceph_cap {
163de57606cSSage Weil 	struct ceph_inode_info *ci;
164de57606cSSage Weil 	struct rb_node ci_node;          /* per-ci cap tree */
165de57606cSSage Weil 	struct ceph_mds_session *session;
166de57606cSSage Weil 	struct list_head session_caps;   /* per-session caplist */
167de57606cSSage Weil 	int mds;
168de57606cSSage Weil 	u64 cap_id;       /* unique cap id (mds provided) */
169de57606cSSage Weil 	int issued;       /* latest, from the mds */
170de57606cSSage Weil 	int implemented;  /* implemented superset of issued (for revocation) */
171de57606cSSage Weil 	int mds_wanted;
172de57606cSSage Weil 	u32 seq, issue_seq, mseq, gen;
173de57606cSSage Weil 	unsigned long last_used;
174de57606cSSage Weil 	struct list_head caps_item;
175de57606cSSage Weil };
176de57606cSSage Weil 
177de57606cSSage Weil #define CHECK_CAPS_NODELAY    1  /* do not delay any further */
178de57606cSSage Weil #define CHECK_CAPS_AUTHONLY   2  /* only check auth cap */
179de57606cSSage Weil #define CHECK_CAPS_FLUSH      4  /* flush any dirty caps */
180de57606cSSage Weil 
181de57606cSSage Weil /*
182de57606cSSage Weil  * Snapped cap state that is pending flush to mds.  When a snapshot occurs,
183de57606cSSage Weil  * we first complete any in-process sync writes and writeback any dirty
184de57606cSSage Weil  * data before flushing the snapped state (tracked here) back to the MDS.
185de57606cSSage Weil  */
186de57606cSSage Weil struct ceph_cap_snap {
187de57606cSSage Weil 	atomic_t nref;
188de57606cSSage Weil 	struct ceph_inode_info *ci;
189de57606cSSage Weil 	struct list_head ci_item, flushing_item;
190de57606cSSage Weil 
191de57606cSSage Weil 	u64 follows, flush_tid;
192de57606cSSage Weil 	int issued, dirty;
193de57606cSSage Weil 	struct ceph_snap_context *context;
194de57606cSSage Weil 
195de57606cSSage Weil 	mode_t mode;
196de57606cSSage Weil 	uid_t uid;
197de57606cSSage Weil 	gid_t gid;
198de57606cSSage Weil 
199de57606cSSage Weil 	void *xattr_blob;
200de57606cSSage Weil 	int xattr_len;
201de57606cSSage Weil 	u64 xattr_version;
202de57606cSSage Weil 
203de57606cSSage Weil 	u64 size;
204de57606cSSage Weil 	struct timespec mtime, atime, ctime;
205de57606cSSage Weil 	u64 time_warp_seq;
206de57606cSSage Weil 	int writing;   /* a sync write is still in progress */
207de57606cSSage Weil 	int dirty_pages;     /* dirty pages awaiting writeback */
208de57606cSSage Weil };
209de57606cSSage Weil 
210de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
211de57606cSSage Weil {
212de57606cSSage Weil 	if (atomic_dec_and_test(&capsnap->nref))
213de57606cSSage Weil 		kfree(capsnap);
214de57606cSSage Weil }
215de57606cSSage Weil 
216de57606cSSage Weil /*
217de57606cSSage Weil  * The frag tree describes how a directory is fragmented, potentially across
218de57606cSSage Weil  * multiple metadata servers.  It is also used to indicate points where
219de57606cSSage Weil  * metadata authority is delegated, and whether/where metadata is replicated.
220de57606cSSage Weil  *
221de57606cSSage Weil  * A _leaf_ frag will be present in the i_fragtree IFF there is
222de57606cSSage Weil  * delegation info.  That is, if mds >= 0 || ndist > 0.
223de57606cSSage Weil  */
224de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4
225de57606cSSage Weil 
226de57606cSSage Weil struct ceph_inode_frag {
227de57606cSSage Weil 	struct rb_node node;
228de57606cSSage Weil 
229de57606cSSage Weil 	/* fragtree state */
230de57606cSSage Weil 	u32 frag;
231de57606cSSage Weil 	int split_by;         /* i.e. 2^(split_by) children */
232de57606cSSage Weil 
233de57606cSSage Weil 	/* delegation and replication info */
234de57606cSSage Weil 	int mds;              /* -1 if same authority as parent */
235de57606cSSage Weil 	int ndist;            /* >0 if replicated */
236de57606cSSage Weil 	int dist[CEPH_MAX_DIRFRAG_REP];
237de57606cSSage Weil };
238de57606cSSage Weil 
239de57606cSSage Weil /*
240de57606cSSage Weil  * We cache inode xattrs as an encoded blob until they are first used,
241de57606cSSage Weil  * at which point we parse them into an rbtree.
242de57606cSSage Weil  */
243de57606cSSage Weil struct ceph_inode_xattr {
244de57606cSSage Weil 	struct rb_node node;
245de57606cSSage Weil 
246de57606cSSage Weil 	const char *name;
247de57606cSSage Weil 	int name_len;
248de57606cSSage Weil 	const char *val;
249de57606cSSage Weil 	int val_len;
250de57606cSSage Weil 	int dirty;
251de57606cSSage Weil 
252de57606cSSage Weil 	int should_free_name;
253de57606cSSage Weil 	int should_free_val;
254de57606cSSage Weil };
255de57606cSSage Weil 
256de57606cSSage Weil struct ceph_inode_xattrs_info {
257de57606cSSage Weil 	/*
258de57606cSSage Weil 	 * (still encoded) xattr blob. we avoid the overhead of parsing
259de57606cSSage Weil 	 * this until someone actually calls getxattr, etc.
260de57606cSSage Weil 	 *
261de57606cSSage Weil 	 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
262de57606cSSage Weil 	 * NULL means we don't know.
263de57606cSSage Weil 	*/
264de57606cSSage Weil 	struct ceph_buffer *blob, *prealloc_blob;
265de57606cSSage Weil 
266de57606cSSage Weil 	struct rb_root index;
267de57606cSSage Weil 	bool dirty;
268de57606cSSage Weil 	int count;
269de57606cSSage Weil 	int names_size;
270de57606cSSage Weil 	int vals_size;
271de57606cSSage Weil 	u64 version, index_version;
272de57606cSSage Weil };
273de57606cSSage Weil 
274de57606cSSage Weil /*
275de57606cSSage Weil  * Ceph inode.
276de57606cSSage Weil  */
277de57606cSSage Weil #define CEPH_I_COMPLETE  1  /* we have complete directory cached */
278de57606cSSage Weil #define CEPH_I_NODELAY   4  /* do not delay cap release */
279de57606cSSage Weil #define CEPH_I_FLUSH     8  /* do not delay flush of dirty metadata */
280de57606cSSage Weil 
281de57606cSSage Weil struct ceph_inode_info {
282de57606cSSage Weil 	struct ceph_vino i_vino;   /* ceph ino + snap */
283de57606cSSage Weil 
284de57606cSSage Weil 	u64 i_version;
285de57606cSSage Weil 	u32 i_time_warp_seq;
286de57606cSSage Weil 
287de57606cSSage Weil 	unsigned i_ceph_flags;
288de57606cSSage Weil 	unsigned long i_release_count;
289de57606cSSage Weil 
290de57606cSSage Weil 	struct ceph_file_layout i_layout;
291de57606cSSage Weil 	char *i_symlink;
292de57606cSSage Weil 
293de57606cSSage Weil 	/* for dirs */
294de57606cSSage Weil 	struct timespec i_rctime;
295de57606cSSage Weil 	u64 i_rbytes, i_rfiles, i_rsubdirs;
296de57606cSSage Weil 	u64 i_files, i_subdirs;
297de57606cSSage Weil 	u64 i_max_offset;  /* largest readdir offset, set with I_COMPLETE */
298de57606cSSage Weil 
299de57606cSSage Weil 	struct rb_root i_fragtree;
300de57606cSSage Weil 	struct mutex i_fragtree_mutex;
301de57606cSSage Weil 
302de57606cSSage Weil 	struct ceph_inode_xattrs_info i_xattrs;
303de57606cSSage Weil 
304de57606cSSage Weil 	/* capabilities.  protected _both_ by i_lock and cap->session's
305de57606cSSage Weil 	 * s_mutex. */
306de57606cSSage Weil 	struct rb_root i_caps;           /* cap list */
307de57606cSSage Weil 	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
308de57606cSSage Weil 	unsigned i_dirty_caps, i_flushing_caps;     /* mask of dirtied fields */
309de57606cSSage Weil 	struct list_head i_dirty_item, i_flushing_item;
310de57606cSSage Weil 	u64 i_cap_flush_seq;
311de57606cSSage Weil 	/* we need to track cap writeback on a per-cap-bit basis, to allow
312de57606cSSage Weil 	 * overlapping, pipelined cap flushes to the mds.  we can probably
313de57606cSSage Weil 	 * reduce the tid to 8 bits if we're concerned about inode size. */
314de57606cSSage Weil 	u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS];
315de57606cSSage Weil 	wait_queue_head_t i_cap_wq;      /* threads waiting on a capability */
316de57606cSSage Weil 	unsigned long i_hold_caps_min; /* jiffies */
317de57606cSSage Weil 	unsigned long i_hold_caps_max; /* jiffies */
318de57606cSSage Weil 	struct list_head i_cap_delay_list;  /* for delayed cap release to mds */
319de57606cSSage Weil 	int i_cap_exporting_mds;         /* to handle cap migration between */
320de57606cSSage Weil 	unsigned i_cap_exporting_mseq;   /*  mds's. */
321de57606cSSage Weil 	unsigned i_cap_exporting_issued;
322de57606cSSage Weil 	struct ceph_cap_reservation i_cap_migration_resv;
323de57606cSSage Weil 	struct list_head i_cap_snaps;   /* snapped state pending flush to mds */
324de57606cSSage Weil 	struct ceph_snap_context *i_head_snapc;  /* set if wr_buffer_head > 0 */
325de57606cSSage Weil 	unsigned i_snap_caps;           /* cap bits for snapped files */
326de57606cSSage Weil 
327de57606cSSage Weil 	int i_nr_by_mode[CEPH_FILE_MODE_NUM];  /* open file counts */
328de57606cSSage Weil 
329de57606cSSage Weil 	u32 i_truncate_seq;        /* last truncate to smaller size */
330de57606cSSage Weil 	u64 i_truncate_size;       /*  and the size we last truncated down to */
331de57606cSSage Weil 	int i_truncate_pending;    /*  still need to call vmtruncate */
332de57606cSSage Weil 
333de57606cSSage Weil 	u64 i_max_size;            /* max file size authorized by mds */
334de57606cSSage Weil 	u64 i_reported_size; /* (max_)size reported to or requested of mds */
335de57606cSSage Weil 	u64 i_wanted_max_size;     /* offset we'd like to write too */
336de57606cSSage Weil 	u64 i_requested_max_size;  /* max_size we've requested */
337de57606cSSage Weil 
338de57606cSSage Weil 	/* held references to caps */
339de57606cSSage Weil 	int i_pin_ref;
340de57606cSSage Weil 	int i_rd_ref, i_rdcache_ref, i_wr_ref;
341de57606cSSage Weil 	int i_wrbuffer_ref, i_wrbuffer_ref_head;
342de57606cSSage Weil 	u32 i_shared_gen;       /* increment each time we get FILE_SHARED */
343de57606cSSage Weil 	u32 i_rdcache_gen;      /* we increment this each time we get
344de57606cSSage Weil 				   FILE_CACHE.  If it's non-zero, we
345de57606cSSage Weil 				   _may_ have cached pages. */
346de57606cSSage Weil 	u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
347de57606cSSage Weil 
348de57606cSSage Weil 	struct list_head i_unsafe_writes; /* uncommitted sync writes */
349de57606cSSage Weil 	struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
350de57606cSSage Weil 	spinlock_t i_unsafe_lock;
351de57606cSSage Weil 
352de57606cSSage Weil 	struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
353de57606cSSage Weil 	int i_snap_realm_counter; /* snap realm (if caps) */
354de57606cSSage Weil 	struct list_head i_snap_realm_item;
355de57606cSSage Weil 	struct list_head i_snap_flush_item;
356de57606cSSage Weil 
357de57606cSSage Weil 	struct work_struct i_wb_work;  /* writeback work */
358de57606cSSage Weil 	struct work_struct i_pg_inv_work;  /* page invalidation work */
359de57606cSSage Weil 
360de57606cSSage Weil 	struct work_struct i_vmtruncate_work;
361de57606cSSage Weil 
362de57606cSSage Weil 	struct inode vfs_inode; /* at end */
363de57606cSSage Weil };
364de57606cSSage Weil 
365de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
366de57606cSSage Weil {
367de57606cSSage Weil 	return list_entry(inode, struct ceph_inode_info, vfs_inode);
368de57606cSSage Weil }
369de57606cSSage Weil 
370de57606cSSage Weil static inline void ceph_i_clear(struct inode *inode, unsigned mask)
371de57606cSSage Weil {
372de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
373de57606cSSage Weil 
374de57606cSSage Weil 	spin_lock(&inode->i_lock);
375de57606cSSage Weil 	ci->i_ceph_flags &= ~mask;
376de57606cSSage Weil 	spin_unlock(&inode->i_lock);
377de57606cSSage Weil }
378de57606cSSage Weil 
379de57606cSSage Weil static inline void ceph_i_set(struct inode *inode, unsigned mask)
380de57606cSSage Weil {
381de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
382de57606cSSage Weil 
383de57606cSSage Weil 	spin_lock(&inode->i_lock);
384de57606cSSage Weil 	ci->i_ceph_flags |= mask;
385de57606cSSage Weil 	spin_unlock(&inode->i_lock);
386de57606cSSage Weil }
387de57606cSSage Weil 
388de57606cSSage Weil static inline bool ceph_i_test(struct inode *inode, unsigned mask)
389de57606cSSage Weil {
390de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
391de57606cSSage Weil 	bool r;
392de57606cSSage Weil 
393de57606cSSage Weil 	smp_mb();
394de57606cSSage Weil 	r = (ci->i_ceph_flags & mask) == mask;
395de57606cSSage Weil 	return r;
396de57606cSSage Weil }
397de57606cSSage Weil 
398de57606cSSage Weil 
399de57606cSSage Weil /* find a specific frag @f */
400de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
401de57606cSSage Weil 						u32 f);
402de57606cSSage Weil 
403de57606cSSage Weil /*
404de57606cSSage Weil  * choose fragment for value @v.  copy frag content to pfrag, if leaf
405de57606cSSage Weil  * exists
406de57606cSSage Weil  */
407de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
408de57606cSSage Weil 			    struct ceph_inode_frag *pfrag,
409de57606cSSage Weil 			    int *found);
410de57606cSSage Weil 
411de57606cSSage Weil /*
412de57606cSSage Weil  * Ceph dentry state
413de57606cSSage Weil  */
414de57606cSSage Weil struct ceph_dentry_info {
415de57606cSSage Weil 	struct ceph_mds_session *lease_session;
416de57606cSSage Weil 	u32 lease_gen, lease_shared_gen;
417de57606cSSage Weil 	u32 lease_seq;
418de57606cSSage Weil 	unsigned long lease_renew_after, lease_renew_from;
419de57606cSSage Weil 	struct list_head lru;
420de57606cSSage Weil 	struct dentry *dentry;
421de57606cSSage Weil 	u64 time;
422de57606cSSage Weil 	u64 offset;
423de57606cSSage Weil };
424de57606cSSage Weil 
425de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
426de57606cSSage Weil {
427de57606cSSage Weil 	return (struct ceph_dentry_info *)dentry->d_fsdata;
428de57606cSSage Weil }
429de57606cSSage Weil 
430de57606cSSage Weil static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
431de57606cSSage Weil {
432de57606cSSage Weil 	return ((loff_t)frag << 32) | (loff_t)off;
433de57606cSSage Weil }
434de57606cSSage Weil 
435de57606cSSage Weil /*
436de57606cSSage Weil  * ino_t is <64 bits on many architectures, blech.
437de57606cSSage Weil  *
438de57606cSSage Weil  * don't include snap in ino hash, at least for now.
439de57606cSSage Weil  */
440de57606cSSage Weil static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
441de57606cSSage Weil {
442de57606cSSage Weil 	ino_t ino = (ino_t)vino.ino;  /* ^ (vino.snap << 20); */
443de57606cSSage Weil #if BITS_PER_LONG == 32
444de57606cSSage Weil 	ino ^= vino.ino >> (sizeof(u64)-sizeof(ino_t)) * 8;
445de57606cSSage Weil 	if (!ino)
446de57606cSSage Weil 		ino = 1;
447de57606cSSage Weil #endif
448de57606cSSage Weil 	return ino;
449de57606cSSage Weil }
450de57606cSSage Weil 
451de57606cSSage Weil static inline int ceph_set_ino_cb(struct inode *inode, void *data)
452de57606cSSage Weil {
453de57606cSSage Weil 	ceph_inode(inode)->i_vino = *(struct ceph_vino *)data;
454de57606cSSage Weil 	inode->i_ino = ceph_vino_to_ino(*(struct ceph_vino *)data);
455de57606cSSage Weil 	return 0;
456de57606cSSage Weil }
457de57606cSSage Weil 
458de57606cSSage Weil static inline struct ceph_vino ceph_vino(struct inode *inode)
459de57606cSSage Weil {
460de57606cSSage Weil 	return ceph_inode(inode)->i_vino;
461de57606cSSage Weil }
462de57606cSSage Weil 
463de57606cSSage Weil /* for printf-style formatting */
464de57606cSSage Weil #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
465de57606cSSage Weil 
466de57606cSSage Weil static inline u64 ceph_ino(struct inode *inode)
467de57606cSSage Weil {
468de57606cSSage Weil 	return ceph_inode(inode)->i_vino.ino;
469de57606cSSage Weil }
470de57606cSSage Weil static inline u64 ceph_snap(struct inode *inode)
471de57606cSSage Weil {
472de57606cSSage Weil 	return ceph_inode(inode)->i_vino.snap;
473de57606cSSage Weil }
474de57606cSSage Weil 
475de57606cSSage Weil static inline int ceph_ino_compare(struct inode *inode, void *data)
476de57606cSSage Weil {
477de57606cSSage Weil 	struct ceph_vino *pvino = (struct ceph_vino *)data;
478de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
479de57606cSSage Weil 	return ci->i_vino.ino == pvino->ino &&
480de57606cSSage Weil 		ci->i_vino.snap == pvino->snap;
481de57606cSSage Weil }
482de57606cSSage Weil 
483de57606cSSage Weil static inline struct inode *ceph_find_inode(struct super_block *sb,
484de57606cSSage Weil 					    struct ceph_vino vino)
485de57606cSSage Weil {
486de57606cSSage Weil 	ino_t t = ceph_vino_to_ino(vino);
487de57606cSSage Weil 	return ilookup5(sb, t, ceph_ino_compare, &vino);
488de57606cSSage Weil }
489de57606cSSage Weil 
490de57606cSSage Weil 
491de57606cSSage Weil /*
492de57606cSSage Weil  * caps helpers
493de57606cSSage Weil  */
494de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
495de57606cSSage Weil {
496de57606cSSage Weil 	return !RB_EMPTY_ROOT(&ci->i_caps);
497de57606cSSage Weil }
498de57606cSSage Weil 
499de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
500de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
501de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
502de57606cSSage Weil 				    struct ceph_cap *cap);
503de57606cSSage Weil 
504de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci)
505de57606cSSage Weil {
506de57606cSSage Weil 	int issued;
507de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
508de57606cSSage Weil 	issued = __ceph_caps_issued(ci, NULL);
509de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
510de57606cSSage Weil 	return issued;
511de57606cSSage Weil }
512de57606cSSage Weil 
513de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
514de57606cSSage Weil 					int touch)
515de57606cSSage Weil {
516de57606cSSage Weil 	int r;
517de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
518de57606cSSage Weil 	r = __ceph_caps_issued_mask(ci, mask, touch);
519de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
520de57606cSSage Weil 	return r;
521de57606cSSage Weil }
522de57606cSSage Weil 
523de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
524de57606cSSage Weil {
525de57606cSSage Weil 	return ci->i_dirty_caps | ci->i_flushing_caps;
526de57606cSSage Weil }
527afcdaea3SSage Weil extern void __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask);
528de57606cSSage Weil 
529de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
530de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci);
531de57606cSSage Weil 
532de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
533de57606cSSage Weil 
534de57606cSSage Weil /*
535de57606cSSage Weil  * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
536de57606cSSage Weil  */
537de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
538de57606cSSage Weil {
539de57606cSSage Weil 	int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
540de57606cSSage Weil 	if (w & CEPH_CAP_FILE_BUFFER)
541de57606cSSage Weil 		w |= CEPH_CAP_FILE_EXCL;  /* we want EXCL if dirty data */
542de57606cSSage Weil 	return w;
543de57606cSSage Weil }
544de57606cSSage Weil 
545de57606cSSage Weil /* what the mds thinks we want */
546de57606cSSage Weil extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
547de57606cSSage Weil 
548de57606cSSage Weil extern void ceph_caps_init(void);
549de57606cSSage Weil extern void ceph_caps_finalize(void);
550de57606cSSage Weil extern int ceph_reserve_caps(struct ceph_cap_reservation *ctx, int need);
551de57606cSSage Weil extern int ceph_unreserve_caps(struct ceph_cap_reservation *ctx);
552de57606cSSage Weil extern void ceph_reservation_status(struct ceph_client *client,
553de57606cSSage Weil 				    int *total, int *avail, int *used,
554de57606cSSage Weil 				    int *reserved);
555de57606cSSage Weil 
556de57606cSSage Weil static inline struct ceph_client *ceph_inode_to_client(struct inode *inode)
557de57606cSSage Weil {
558de57606cSSage Weil 	return (struct ceph_client *)inode->i_sb->s_fs_info;
559de57606cSSage Weil }
560de57606cSSage Weil 
561de57606cSSage Weil static inline struct ceph_client *ceph_sb_to_client(struct super_block *sb)
562de57606cSSage Weil {
563de57606cSSage Weil 	return (struct ceph_client *)sb->s_fs_info;
564de57606cSSage Weil }
565de57606cSSage Weil 
566de57606cSSage Weil static inline int ceph_queue_writeback(struct inode *inode)
567de57606cSSage Weil {
568de57606cSSage Weil 	return queue_work(ceph_inode_to_client(inode)->wb_wq,
569de57606cSSage Weil 		   &ceph_inode(inode)->i_wb_work);
570de57606cSSage Weil }
571de57606cSSage Weil 
572de57606cSSage Weil static inline int ceph_queue_page_invalidation(struct inode *inode)
573de57606cSSage Weil {
574de57606cSSage Weil 	return queue_work(ceph_inode_to_client(inode)->pg_inv_wq,
575de57606cSSage Weil 		   &ceph_inode(inode)->i_pg_inv_work);
576de57606cSSage Weil }
577de57606cSSage Weil 
578de57606cSSage Weil 
579de57606cSSage Weil /*
580de57606cSSage Weil  * we keep buffered readdir results attached to file->private_data
581de57606cSSage Weil  */
582de57606cSSage Weil struct ceph_file_info {
583de57606cSSage Weil 	int fmode;     /* initialized on open */
584de57606cSSage Weil 
585de57606cSSage Weil 	/* readdir: position within the dir */
586de57606cSSage Weil 	u32 frag;
587de57606cSSage Weil 	struct ceph_mds_request *last_readdir;
588de57606cSSage Weil 	int at_end;
589de57606cSSage Weil 
590de57606cSSage Weil 	/* readdir: position within a frag */
591de57606cSSage Weil 	unsigned offset;       /* offset of last chunk, adjusted for . and .. */
592de57606cSSage Weil 	u64 next_offset;       /* offset of next chunk (last_name's + 1) */
593de57606cSSage Weil 	char *last_name;       /* last entry in previous chunk */
594de57606cSSage Weil 	struct dentry *dentry; /* next dentry (for dcache readdir) */
595de57606cSSage Weil 	unsigned long dir_release_count;
596de57606cSSage Weil 
597de57606cSSage Weil 	/* used for -o dirstat read() on directory thing */
598de57606cSSage Weil 	char *dir_info;
599de57606cSSage Weil 	int dir_info_len;
600de57606cSSage Weil };
601de57606cSSage Weil 
602de57606cSSage Weil 
603de57606cSSage Weil 
604de57606cSSage Weil /*
605de57606cSSage Weil  * snapshots
606de57606cSSage Weil  */
607de57606cSSage Weil 
608de57606cSSage Weil /*
609de57606cSSage Weil  * A "snap context" is the set of existing snapshots when we
610de57606cSSage Weil  * write data.  It is used by the OSD to guide its COW behavior.
611de57606cSSage Weil  *
612de57606cSSage Weil  * The ceph_snap_context is refcounted, and attached to each dirty
613de57606cSSage Weil  * page, indicating which context the dirty data belonged when it was
614de57606cSSage Weil  * dirtied.
615de57606cSSage Weil  */
616de57606cSSage Weil struct ceph_snap_context {
617de57606cSSage Weil 	atomic_t nref;
618de57606cSSage Weil 	u64 seq;
619de57606cSSage Weil 	int num_snaps;
620de57606cSSage Weil 	u64 snaps[];
621de57606cSSage Weil };
622de57606cSSage Weil 
623de57606cSSage Weil static inline struct ceph_snap_context *
624de57606cSSage Weil ceph_get_snap_context(struct ceph_snap_context *sc)
625de57606cSSage Weil {
626de57606cSSage Weil 	/*
627de57606cSSage Weil 	printk("get_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
628de57606cSSage Weil 	       atomic_read(&sc->nref)+1);
629de57606cSSage Weil 	*/
630de57606cSSage Weil 	if (sc)
631de57606cSSage Weil 		atomic_inc(&sc->nref);
632de57606cSSage Weil 	return sc;
633de57606cSSage Weil }
634de57606cSSage Weil 
635de57606cSSage Weil static inline void ceph_put_snap_context(struct ceph_snap_context *sc)
636de57606cSSage Weil {
637de57606cSSage Weil 	if (!sc)
638de57606cSSage Weil 		return;
639de57606cSSage Weil 	/*
640de57606cSSage Weil 	printk("put_snap_context %p %d -> %d\n", sc, atomic_read(&sc->nref),
641de57606cSSage Weil 	       atomic_read(&sc->nref)-1);
642de57606cSSage Weil 	*/
643de57606cSSage Weil 	if (atomic_dec_and_test(&sc->nref)) {
644de57606cSSage Weil 		/*printk(" deleting snap_context %p\n", sc);*/
645de57606cSSage Weil 		kfree(sc);
646de57606cSSage Weil 	}
647de57606cSSage Weil }
648de57606cSSage Weil 
649de57606cSSage Weil /*
650de57606cSSage Weil  * A "snap realm" describes a subset of the file hierarchy sharing
651de57606cSSage Weil  * the same set of snapshots that apply to it.  The realms themselves
652de57606cSSage Weil  * are organized into a hierarchy, such that children inherit (some of)
653de57606cSSage Weil  * the snapshots of their parents.
654de57606cSSage Weil  *
655de57606cSSage Weil  * All inodes within the realm that have capabilities are linked into a
656de57606cSSage Weil  * per-realm list.
657de57606cSSage Weil  */
658de57606cSSage Weil struct ceph_snap_realm {
659de57606cSSage Weil 	u64 ino;
660de57606cSSage Weil 	atomic_t nref;
661de57606cSSage Weil 	u64 created, seq;
662de57606cSSage Weil 	u64 parent_ino;
663de57606cSSage Weil 	u64 parent_since;   /* snapid when our current parent became so */
664de57606cSSage Weil 
665de57606cSSage Weil 	u64 *prior_parent_snaps;      /* snaps inherited from any parents we */
666de57606cSSage Weil 	int num_prior_parent_snaps;   /*  had prior to parent_since */
667de57606cSSage Weil 	u64 *snaps;                   /* snaps specific to this realm */
668de57606cSSage Weil 	int num_snaps;
669de57606cSSage Weil 
670de57606cSSage Weil 	struct ceph_snap_realm *parent;
671de57606cSSage Weil 	struct list_head children;       /* list of child realms */
672de57606cSSage Weil 	struct list_head child_item;
673de57606cSSage Weil 
674de57606cSSage Weil 	struct list_head empty_item;     /* if i have ref==0 */
675de57606cSSage Weil 
676de57606cSSage Weil 	/* the current set of snaps for this realm */
677de57606cSSage Weil 	struct ceph_snap_context *cached_context;
678de57606cSSage Weil 
679de57606cSSage Weil 	struct list_head inodes_with_caps;
680de57606cSSage Weil 	spinlock_t inodes_with_caps_lock;
681de57606cSSage Weil };
682de57606cSSage Weil 
683de57606cSSage Weil 
684de57606cSSage Weil 
685de57606cSSage Weil /*
686de57606cSSage Weil  * calculate the number of pages a given length and offset map onto,
687de57606cSSage Weil  * if we align the data.
688de57606cSSage Weil  */
689de57606cSSage Weil static inline int calc_pages_for(u64 off, u64 len)
690de57606cSSage Weil {
691de57606cSSage Weil 	return ((off+len+PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT) -
692de57606cSSage Weil 		(off >> PAGE_CACHE_SHIFT);
693de57606cSSage Weil }
694de57606cSSage Weil 
695de57606cSSage Weil 
696de57606cSSage Weil 
697de57606cSSage Weil /* snap.c */
698de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
699de57606cSSage Weil 					       u64 ino);
700de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
701de57606cSSage Weil 				struct ceph_snap_realm *realm);
702de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
703de57606cSSage Weil 				struct ceph_snap_realm *realm);
704de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m,
705de57606cSSage Weil 				  void *p, void *e, bool deletion);
706de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
707de57606cSSage Weil 			     struct ceph_msg *msg);
708de57606cSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci,
709de57606cSSage Weil 				struct ceph_snap_context *snapc);
710de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
711de57606cSSage Weil 				  struct ceph_cap_snap *capsnap);
712de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
713de57606cSSage Weil 
714de57606cSSage Weil /*
715de57606cSSage Weil  * a cap_snap is "pending" if it is still awaiting an in-progress
716de57606cSSage Weil  * sync write (that may/may not still update size, mtime, etc.).
717de57606cSSage Weil  */
718de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
719de57606cSSage Weil {
720de57606cSSage Weil 	return !list_empty(&ci->i_cap_snaps) &&
721de57606cSSage Weil 		list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap,
722de57606cSSage Weil 			   ci_item)->writing;
723de57606cSSage Weil }
724de57606cSSage Weil 
725de57606cSSage Weil 
726de57606cSSage Weil /* super.c */
727de57606cSSage Weil extern struct kmem_cache *ceph_inode_cachep;
728de57606cSSage Weil extern struct kmem_cache *ceph_cap_cachep;
729de57606cSSage Weil extern struct kmem_cache *ceph_dentry_cachep;
730de57606cSSage Weil extern struct kmem_cache *ceph_file_cachep;
731de57606cSSage Weil 
732de57606cSSage Weil extern const char *ceph_msg_type_name(int type);
733de57606cSSage Weil 
734de57606cSSage Weil #define FSID_FORMAT "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-" \
735de57606cSSage Weil 	"%02x%02x%02x%02x%02x%02x"
736de57606cSSage Weil #define PR_FSID(f) (f)->fsid[0], (f)->fsid[1], (f)->fsid[2], (f)->fsid[3], \
737de57606cSSage Weil 		(f)->fsid[4], (f)->fsid[5], (f)->fsid[6], (f)->fsid[7],    \
738de57606cSSage Weil 		(f)->fsid[8], (f)->fsid[9], (f)->fsid[10], (f)->fsid[11],  \
739de57606cSSage Weil 		(f)->fsid[12], (f)->fsid[13], (f)->fsid[14], (f)->fsid[15]
740de57606cSSage Weil 
741de57606cSSage Weil /* inode.c */
742de57606cSSage Weil extern const struct inode_operations ceph_file_iops;
743de57606cSSage Weil 
744de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb);
745de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode);
746de57606cSSage Weil 
747de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb,
748de57606cSSage Weil 				    struct ceph_vino vino);
749de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent);
750de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued,
751de57606cSSage Weil 			       u32 truncate_seq, u64 truncate_size, u64 size);
752de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued,
753de57606cSSage Weil 				u64 time_warp_seq, struct timespec *ctime,
754de57606cSSage Weil 				struct timespec *mtime, struct timespec *atime);
755de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb,
756de57606cSSage Weil 			   struct ceph_mds_request *req,
757de57606cSSage Weil 			   struct ceph_mds_session *session);
758de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
759de57606cSSage Weil 				    struct ceph_mds_session *session);
760de57606cSSage Weil 
761de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask);
762de57606cSSage Weil 
763de57606cSSage Weil extern int ceph_inode_set_size(struct inode *inode, loff_t size);
764de57606cSSage Weil extern void ceph_inode_writeback(struct work_struct *work);
765de57606cSSage Weil extern void ceph_vmtruncate_work(struct work_struct *work);
766de57606cSSage Weil extern void __ceph_do_pending_vmtruncate(struct inode *inode);
767de57606cSSage Weil extern void __ceph_queue_vmtruncate(struct inode *inode);
768de57606cSSage Weil 
769de57606cSSage Weil extern int ceph_do_getattr(struct inode *inode, int mask);
770de57606cSSage Weil extern int ceph_permission(struct inode *inode, int mask);
771de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
772de57606cSSage Weil extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
773de57606cSSage Weil 			struct kstat *stat);
774de57606cSSage Weil 
775de57606cSSage Weil /* xattr.c */
776de57606cSSage Weil extern int ceph_setxattr(struct dentry *, const char *, const void *,
777de57606cSSage Weil 			 size_t, int);
778de57606cSSage Weil extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t);
779de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
780de57606cSSage Weil extern int ceph_removexattr(struct dentry *, const char *);
781de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
782de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
783de57606cSSage Weil 
784de57606cSSage Weil /* caps.c */
785de57606cSSage Weil extern const char *ceph_cap_string(int c);
786de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session,
787de57606cSSage Weil 			     struct ceph_msg *msg);
788de57606cSSage Weil extern int ceph_add_cap(struct inode *inode,
789de57606cSSage Weil 			struct ceph_mds_session *session, u64 cap_id,
790de57606cSSage Weil 			int fmode, unsigned issued, unsigned wanted,
791de57606cSSage Weil 			unsigned cap, unsigned seq, u64 realmino, int flags,
792de57606cSSage Weil 			struct ceph_cap_reservation *caps_reservation);
793de57606cSSage Weil extern void __ceph_remove_cap(struct ceph_cap *cap,
794de57606cSSage Weil 			      struct ceph_cap_reservation *ctx);
795de57606cSSage Weil static inline void ceph_remove_cap(struct ceph_cap *cap)
796de57606cSSage Weil {
797de57606cSSage Weil 	struct inode *inode = &cap->ci->vfs_inode;
798de57606cSSage Weil 	spin_lock(&inode->i_lock);
799de57606cSSage Weil 	__ceph_remove_cap(cap, NULL);
800de57606cSSage Weil 	spin_unlock(&inode->i_lock);
801de57606cSSage Weil }
802de57606cSSage Weil 
803de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode);
804de57606cSSage Weil extern int ceph_write_inode(struct inode *inode, int unused);
805de57606cSSage Weil extern int ceph_fsync(struct file *file, struct dentry *dentry, int datasync);
806de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
807de57606cSSage Weil 				    struct ceph_mds_session *session);
808de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode);
809de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
810de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
811de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
812de57606cSSage Weil 				       struct ceph_snap_context *snapc);
813de57606cSSage Weil extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
814de57606cSSage Weil 			       struct ceph_mds_session **psession);
815de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
816de57606cSSage Weil 			    struct ceph_mds_session *session);
817afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
818afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
819de57606cSSage Weil 
820de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode,
821de57606cSSage Weil 				     int mds, int drop, int unless, int force);
822de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
823de57606cSSage Weil 				      int mds, int drop, int unless);
824de57606cSSage Weil 
825de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
826de57606cSSage Weil 			 int *got, loff_t endoff);
827de57606cSSage Weil 
828de57606cSSage Weil /* for counting open files by mode */
829de57606cSSage Weil static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
830de57606cSSage Weil {
831de57606cSSage Weil 	ci->i_nr_by_mode[mode]++;
832de57606cSSage Weil }
833de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
834de57606cSSage Weil 
835de57606cSSage Weil /* addr.c */
836de57606cSSage Weil extern const struct address_space_operations ceph_aops;
837de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
838de57606cSSage Weil 
839de57606cSSage Weil /* file.c */
840de57606cSSage Weil extern const struct file_operations ceph_file_fops;
841de57606cSSage Weil extern const struct address_space_operations ceph_aops;
842de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file);
843de57606cSSage Weil extern struct dentry *ceph_lookup_open(struct inode *dir, struct dentry *dentry,
844de57606cSSage Weil 				       struct nameidata *nd, int mode,
845de57606cSSage Weil 				       int locked_dir);
846de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp);
847de57606cSSage Weil extern void ceph_release_page_vector(struct page **pages, int num_pages);
848de57606cSSage Weil 
849de57606cSSage Weil /* dir.c */
850de57606cSSage Weil extern const struct file_operations ceph_dir_fops;
851de57606cSSage Weil extern const struct inode_operations ceph_dir_iops;
852de57606cSSage Weil extern struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
853de57606cSSage Weil 	ceph_snapdir_dentry_ops;
854de57606cSSage Weil 
855de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
856de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
857de57606cSSage Weil 					 struct dentry *dentry, int err);
858de57606cSSage Weil 
859de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn);
860de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn);
861de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn);
862de57606cSSage Weil 
863de57606cSSage Weil /*
864de57606cSSage Weil  * our d_ops vary depending on whether the inode is live,
865de57606cSSage Weil  * snapshotted (read-only), or a virtual ".snap" directory.
866de57606cSSage Weil  */
867de57606cSSage Weil int ceph_init_dentry(struct dentry *dentry);
868de57606cSSage Weil 
869de57606cSSage Weil 
870de57606cSSage Weil /* ioctl.c */
871de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
872de57606cSSage Weil 
873de57606cSSage Weil /* export.c */
874de57606cSSage Weil extern const struct export_operations ceph_export_ops;
875de57606cSSage Weil 
876de57606cSSage Weil /* debugfs.c */
877de57606cSSage Weil extern int ceph_debugfs_init(void);
878de57606cSSage Weil extern void ceph_debugfs_cleanup(void);
879de57606cSSage Weil extern int ceph_debugfs_client_init(struct ceph_client *client);
880de57606cSSage Weil extern void ceph_debugfs_client_cleanup(struct ceph_client *client);
881de57606cSSage Weil 
882de57606cSSage Weil static inline struct inode *get_dentry_parent_inode(struct dentry *dentry)
883de57606cSSage Weil {
884de57606cSSage Weil 	if (dentry && dentry->d_parent)
885de57606cSSage Weil 		return dentry->d_parent->d_inode;
886de57606cSSage Weil 
887de57606cSSage Weil 	return NULL;
888de57606cSSage Weil }
889de57606cSSage Weil 
890de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */
891