xref: /linux/fs/ceph/super.h (revision 5f21c96dd5c615341963036ae8f5e4f5227a818d)
1de57606cSSage Weil #ifndef _FS_CEPH_SUPER_H
2de57606cSSage Weil #define _FS_CEPH_SUPER_H
3de57606cSSage Weil 
43d14c5d2SYehuda Sadeh #include <linux/ceph/ceph_debug.h>
5de57606cSSage Weil 
6de57606cSSage Weil #include <asm/unaligned.h>
7de57606cSSage Weil #include <linux/backing-dev.h>
8de57606cSSage Weil #include <linux/completion.h>
9de57606cSSage Weil #include <linux/exportfs.h>
10de57606cSSage Weil #include <linux/fs.h>
11de57606cSSage Weil #include <linux/mempool.h>
12de57606cSSage Weil #include <linux/pagemap.h>
13de57606cSSage Weil #include <linux/wait.h>
14f1a3d572SStephen Rothwell #include <linux/writeback.h>
155a0e3ad6STejun Heo #include <linux/slab.h>
16de57606cSSage Weil 
173d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h>
18de57606cSSage Weil 
19de57606cSSage Weil /* f_type in struct statfs */
20de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400
21de57606cSSage Weil 
22de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate
23de57606cSSage Weil  * large volume sizes on 32-bit machines. */
24de57606cSSage Weil #define CEPH_BLOCK_SHIFT   20  /* 1 MB */
25de57606cSSage Weil #define CEPH_BLOCK         (1 << CEPH_BLOCK_SHIFT)
26de57606cSSage Weil 
273d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT         (1<<4) /* `cat dirname` for stats */
283d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES          (1<<5) /* dir st_bytes = rbytes */
293d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR  (1<<7) /* no dcache readdir */
30ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32           (1<<8) /* 32 bit inos */
316a259382SSage Weil 
323d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DEFAULT    (CEPH_MOUNT_OPT_RBYTES)
33de57606cSSage Weil 
343d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \
353d14c5d2SYehuda Sadeh 	(fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt;
363d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \
373d14c5d2SYehuda Sadeh 	(!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
38de57606cSSage Weil 
3980456f86SSage Weil #define CEPH_RSIZE_DEFAULT             (512*1024) /* readahead */
403d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT        1024
413d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT  (512*1024)
423d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT        ".snap"
43de57606cSSage Weil 
443d14c5d2SYehuda Sadeh struct ceph_mount_options {
45de57606cSSage Weil 	int flags;
463d14c5d2SYehuda Sadeh 	int sb_flags;
473d14c5d2SYehuda Sadeh 
486e19a16eSSage Weil 	int wsize;
496e19a16eSSage Weil 	int rsize;            /* max readahead */
506e19a16eSSage Weil 	int congestion_kb;    /* max writeback in flight */
516e19a16eSSage Weil 	int caps_wanted_delay_min, caps_wanted_delay_max;
526e19a16eSSage Weil 	int cap_release_safety;
5323804d91SSage Weil 	int max_readdir;       /* max readdir result (entires) */
5423804d91SSage Weil 	int max_readdir_bytes; /* max readdir result (bytes) */
553d14c5d2SYehuda Sadeh 
563d14c5d2SYehuda Sadeh 	/*
573d14c5d2SYehuda Sadeh 	 * everything above this point can be memcmp'd; everything below
583d14c5d2SYehuda Sadeh 	 * is handled in compare_mount_options()
593d14c5d2SYehuda Sadeh 	 */
603d14c5d2SYehuda Sadeh 
61de57606cSSage Weil 	char *snapdir_name;   /* default ".snap" */
62de57606cSSage Weil };
63de57606cSSage Weil 
643d14c5d2SYehuda Sadeh struct ceph_fs_client {
65de57606cSSage Weil 	struct super_block *sb;
66de57606cSSage Weil 
673d14c5d2SYehuda Sadeh 	struct ceph_mount_options *mount_options;
683d14c5d2SYehuda Sadeh 	struct ceph_client *client;
693d14c5d2SYehuda Sadeh 
70de57606cSSage Weil 	unsigned long mount_state;
7185ccce43SSage Weil 	int min_caps;                  /* min caps i added */
7285ccce43SSage Weil 
733d14c5d2SYehuda Sadeh 	struct ceph_mds_client *mdsc;
74de57606cSSage Weil 
75de57606cSSage Weil 	/* writeback */
76de57606cSSage Weil 	mempool_t *wb_pagevec_pool;
77de57606cSSage Weil 	struct workqueue_struct *wb_wq;
78de57606cSSage Weil 	struct workqueue_struct *pg_inv_wq;
79de57606cSSage Weil 	struct workqueue_struct *trunc_wq;
802baba250SYehuda Sadeh 	atomic_long_t writeback_count;
81de57606cSSage Weil 
82de57606cSSage Weil 	struct backing_dev_info backing_dev_info;
830743304dSSage Weil 
840743304dSSage Weil #ifdef CONFIG_DEBUG_FS
853d14c5d2SYehuda Sadeh 	struct dentry *debugfs_dentry_lru, *debugfs_caps;
862baba250SYehuda Sadeh 	struct dentry *debugfs_congestion_kb;
8706edf046SSage Weil 	struct dentry *debugfs_bdi;
883d14c5d2SYehuda Sadeh 	struct dentry *debugfs_mdsc, *debugfs_mdsmap;
890743304dSSage Weil #endif
90de57606cSSage Weil };
91de57606cSSage Weil 
923d14c5d2SYehuda Sadeh 
93de57606cSSage Weil /*
94de57606cSSage Weil  * File i/o capability.  This tracks shared state with the metadata
95de57606cSSage Weil  * server that allows us to cache or writeback attributes or to read
96de57606cSSage Weil  * and write data.  For any given inode, we should have one or more
97de57606cSSage Weil  * capabilities, one issued by each metadata server, and our
98de57606cSSage Weil  * cumulative access is the OR of all issued capabilities.
99de57606cSSage Weil  *
100de57606cSSage Weil  * Each cap is referenced by the inode's i_caps rbtree and by per-mds
101de57606cSSage Weil  * session capability lists.
102de57606cSSage Weil  */
103de57606cSSage Weil struct ceph_cap {
104de57606cSSage Weil 	struct ceph_inode_info *ci;
105de57606cSSage Weil 	struct rb_node ci_node;          /* per-ci cap tree */
106de57606cSSage Weil 	struct ceph_mds_session *session;
107de57606cSSage Weil 	struct list_head session_caps;   /* per-session caplist */
108de57606cSSage Weil 	int mds;
109de57606cSSage Weil 	u64 cap_id;       /* unique cap id (mds provided) */
110de57606cSSage Weil 	int issued;       /* latest, from the mds */
111de57606cSSage Weil 	int implemented;  /* implemented superset of issued (for revocation) */
112de57606cSSage Weil 	int mds_wanted;
113685f9a5dSSage Weil 	u32 seq, issue_seq, mseq;
114685f9a5dSSage Weil 	u32 cap_gen;      /* active/stale cycle */
115de57606cSSage Weil 	unsigned long last_used;
116de57606cSSage Weil 	struct list_head caps_item;
117de57606cSSage Weil };
118de57606cSSage Weil 
119de57606cSSage Weil #define CHECK_CAPS_NODELAY    1  /* do not delay any further */
120de57606cSSage Weil #define CHECK_CAPS_AUTHONLY   2  /* only check auth cap */
121de57606cSSage Weil #define CHECK_CAPS_FLUSH      4  /* flush any dirty caps */
122de57606cSSage Weil 
123de57606cSSage Weil /*
124de57606cSSage Weil  * Snapped cap state that is pending flush to mds.  When a snapshot occurs,
125de57606cSSage Weil  * we first complete any in-process sync writes and writeback any dirty
126de57606cSSage Weil  * data before flushing the snapped state (tracked here) back to the MDS.
127de57606cSSage Weil  */
128de57606cSSage Weil struct ceph_cap_snap {
129de57606cSSage Weil 	atomic_t nref;
130de57606cSSage Weil 	struct ceph_inode_info *ci;
131de57606cSSage Weil 	struct list_head ci_item, flushing_item;
132de57606cSSage Weil 
133de57606cSSage Weil 	u64 follows, flush_tid;
134de57606cSSage Weil 	int issued, dirty;
135de57606cSSage Weil 	struct ceph_snap_context *context;
136de57606cSSage Weil 
137de57606cSSage Weil 	mode_t mode;
138de57606cSSage Weil 	uid_t uid;
139de57606cSSage Weil 	gid_t gid;
140de57606cSSage Weil 
1414a625be4SSage Weil 	struct ceph_buffer *xattr_blob;
142de57606cSSage Weil 	u64 xattr_version;
143de57606cSSage Weil 
144de57606cSSage Weil 	u64 size;
145de57606cSSage Weil 	struct timespec mtime, atime, ctime;
146de57606cSSage Weil 	u64 time_warp_seq;
147de57606cSSage Weil 	int writing;   /* a sync write is still in progress */
148de57606cSSage Weil 	int dirty_pages;     /* dirty pages awaiting writeback */
149de57606cSSage Weil };
150de57606cSSage Weil 
151de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
152de57606cSSage Weil {
1534a625be4SSage Weil 	if (atomic_dec_and_test(&capsnap->nref)) {
1544a625be4SSage Weil 		if (capsnap->xattr_blob)
1554a625be4SSage Weil 			ceph_buffer_put(capsnap->xattr_blob);
156de57606cSSage Weil 		kfree(capsnap);
157de57606cSSage Weil 	}
1584a625be4SSage Weil }
159de57606cSSage Weil 
160de57606cSSage Weil /*
161de57606cSSage Weil  * The frag tree describes how a directory is fragmented, potentially across
162de57606cSSage Weil  * multiple metadata servers.  It is also used to indicate points where
163de57606cSSage Weil  * metadata authority is delegated, and whether/where metadata is replicated.
164de57606cSSage Weil  *
165de57606cSSage Weil  * A _leaf_ frag will be present in the i_fragtree IFF there is
166de57606cSSage Weil  * delegation info.  That is, if mds >= 0 || ndist > 0.
167de57606cSSage Weil  */
168de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4
169de57606cSSage Weil 
170de57606cSSage Weil struct ceph_inode_frag {
171de57606cSSage Weil 	struct rb_node node;
172de57606cSSage Weil 
173de57606cSSage Weil 	/* fragtree state */
174de57606cSSage Weil 	u32 frag;
175de57606cSSage Weil 	int split_by;         /* i.e. 2^(split_by) children */
176de57606cSSage Weil 
177de57606cSSage Weil 	/* delegation and replication info */
178de57606cSSage Weil 	int mds;              /* -1 if same authority as parent */
179de57606cSSage Weil 	int ndist;            /* >0 if replicated */
180de57606cSSage Weil 	int dist[CEPH_MAX_DIRFRAG_REP];
181de57606cSSage Weil };
182de57606cSSage Weil 
183de57606cSSage Weil /*
184de57606cSSage Weil  * We cache inode xattrs as an encoded blob until they are first used,
185de57606cSSage Weil  * at which point we parse them into an rbtree.
186de57606cSSage Weil  */
187de57606cSSage Weil struct ceph_inode_xattr {
188de57606cSSage Weil 	struct rb_node node;
189de57606cSSage Weil 
190de57606cSSage Weil 	const char *name;
191de57606cSSage Weil 	int name_len;
192de57606cSSage Weil 	const char *val;
193de57606cSSage Weil 	int val_len;
194de57606cSSage Weil 	int dirty;
195de57606cSSage Weil 
196de57606cSSage Weil 	int should_free_name;
197de57606cSSage Weil 	int should_free_val;
198de57606cSSage Weil };
199de57606cSSage Weil 
2003d14c5d2SYehuda Sadeh /*
2013d14c5d2SYehuda Sadeh  * Ceph dentry state
2023d14c5d2SYehuda Sadeh  */
2033d14c5d2SYehuda Sadeh struct ceph_dentry_info {
2043d14c5d2SYehuda Sadeh 	struct ceph_mds_session *lease_session;
2053d14c5d2SYehuda Sadeh 	u32 lease_gen, lease_shared_gen;
2063d14c5d2SYehuda Sadeh 	u32 lease_seq;
2073d14c5d2SYehuda Sadeh 	unsigned long lease_renew_after, lease_renew_from;
2083d14c5d2SYehuda Sadeh 	struct list_head lru;
2093d14c5d2SYehuda Sadeh 	struct dentry *dentry;
2103d14c5d2SYehuda Sadeh 	u64 time;
2113d14c5d2SYehuda Sadeh 	u64 offset;
2123d14c5d2SYehuda Sadeh };
2133d14c5d2SYehuda Sadeh 
214de57606cSSage Weil struct ceph_inode_xattrs_info {
215de57606cSSage Weil 	/*
216de57606cSSage Weil 	 * (still encoded) xattr blob. we avoid the overhead of parsing
217de57606cSSage Weil 	 * this until someone actually calls getxattr, etc.
218de57606cSSage Weil 	 *
219de57606cSSage Weil 	 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
220de57606cSSage Weil 	 * NULL means we don't know.
221de57606cSSage Weil 	*/
222de57606cSSage Weil 	struct ceph_buffer *blob, *prealloc_blob;
223de57606cSSage Weil 
224de57606cSSage Weil 	struct rb_root index;
225de57606cSSage Weil 	bool dirty;
226de57606cSSage Weil 	int count;
227de57606cSSage Weil 	int names_size;
228de57606cSSage Weil 	int vals_size;
229de57606cSSage Weil 	u64 version, index_version;
230de57606cSSage Weil };
231de57606cSSage Weil 
232de57606cSSage Weil /*
233de57606cSSage Weil  * Ceph inode.
234de57606cSSage Weil  */
235de57606cSSage Weil struct ceph_inode_info {
236de57606cSSage Weil 	struct ceph_vino i_vino;   /* ceph ino + snap */
237de57606cSSage Weil 
238de57606cSSage Weil 	u64 i_version;
239de57606cSSage Weil 	u32 i_time_warp_seq;
240de57606cSSage Weil 
241de57606cSSage Weil 	unsigned i_ceph_flags;
242de57606cSSage Weil 	unsigned long i_release_count;
243de57606cSSage Weil 
2446c0f3af7SSage Weil 	struct ceph_dir_layout i_dir_layout;
245de57606cSSage Weil 	struct ceph_file_layout i_layout;
246de57606cSSage Weil 	char *i_symlink;
247de57606cSSage Weil 
248de57606cSSage Weil 	/* for dirs */
249de57606cSSage Weil 	struct timespec i_rctime;
250de57606cSSage Weil 	u64 i_rbytes, i_rfiles, i_rsubdirs;
251de57606cSSage Weil 	u64 i_files, i_subdirs;
252de57606cSSage Weil 	u64 i_max_offset;  /* largest readdir offset, set with I_COMPLETE */
253de57606cSSage Weil 
254de57606cSSage Weil 	struct rb_root i_fragtree;
255de57606cSSage Weil 	struct mutex i_fragtree_mutex;
256de57606cSSage Weil 
257de57606cSSage Weil 	struct ceph_inode_xattrs_info i_xattrs;
258de57606cSSage Weil 
259de57606cSSage Weil 	/* capabilities.  protected _both_ by i_lock and cap->session's
260de57606cSSage Weil 	 * s_mutex. */
261de57606cSSage Weil 	struct rb_root i_caps;           /* cap list */
262de57606cSSage Weil 	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
263de57606cSSage Weil 	unsigned i_dirty_caps, i_flushing_caps;     /* mask of dirtied fields */
264de57606cSSage Weil 	struct list_head i_dirty_item, i_flushing_item;
265de57606cSSage Weil 	u64 i_cap_flush_seq;
266de57606cSSage Weil 	/* we need to track cap writeback on a per-cap-bit basis, to allow
267de57606cSSage Weil 	 * overlapping, pipelined cap flushes to the mds.  we can probably
268de57606cSSage Weil 	 * reduce the tid to 8 bits if we're concerned about inode size. */
269de57606cSSage Weil 	u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS];
270de57606cSSage Weil 	wait_queue_head_t i_cap_wq;      /* threads waiting on a capability */
271de57606cSSage Weil 	unsigned long i_hold_caps_min; /* jiffies */
272de57606cSSage Weil 	unsigned long i_hold_caps_max; /* jiffies */
273de57606cSSage Weil 	struct list_head i_cap_delay_list;  /* for delayed cap release to mds */
274de57606cSSage Weil 	int i_cap_exporting_mds;         /* to handle cap migration between */
275de57606cSSage Weil 	unsigned i_cap_exporting_mseq;   /*  mds's. */
276de57606cSSage Weil 	unsigned i_cap_exporting_issued;
277de57606cSSage Weil 	struct ceph_cap_reservation i_cap_migration_resv;
278de57606cSSage Weil 	struct list_head i_cap_snaps;   /* snapped state pending flush to mds */
2797d8cb26dSSage Weil 	struct ceph_snap_context *i_head_snapc;  /* set if wr_buffer_head > 0 or
2807d8cb26dSSage Weil 						    dirty|flushing caps */
281de57606cSSage Weil 	unsigned i_snap_caps;           /* cap bits for snapped files */
282de57606cSSage Weil 
283de57606cSSage Weil 	int i_nr_by_mode[CEPH_FILE_MODE_NUM];  /* open file counts */
284de57606cSSage Weil 
285de57606cSSage Weil 	u32 i_truncate_seq;        /* last truncate to smaller size */
286de57606cSSage Weil 	u64 i_truncate_size;       /*  and the size we last truncated down to */
287de57606cSSage Weil 	int i_truncate_pending;    /*  still need to call vmtruncate */
288de57606cSSage Weil 
289de57606cSSage Weil 	u64 i_max_size;            /* max file size authorized by mds */
290de57606cSSage Weil 	u64 i_reported_size; /* (max_)size reported to or requested of mds */
291de57606cSSage Weil 	u64 i_wanted_max_size;     /* offset we'd like to write too */
292de57606cSSage Weil 	u64 i_requested_max_size;  /* max_size we've requested */
293de57606cSSage Weil 
294de57606cSSage Weil 	/* held references to caps */
295de57606cSSage Weil 	int i_pin_ref;
296d3d0720dSHenry C Chang 	int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref;
297de57606cSSage Weil 	int i_wrbuffer_ref, i_wrbuffer_ref_head;
298de57606cSSage Weil 	u32 i_shared_gen;       /* increment each time we get FILE_SHARED */
299cd045cb4SSage Weil 	u32 i_rdcache_gen;      /* incremented each time we get FILE_CACHE. */
300de57606cSSage Weil 	u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
301de57606cSSage Weil 
302de57606cSSage Weil 	struct list_head i_unsafe_writes; /* uncommitted sync writes */
303de57606cSSage Weil 	struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
304de57606cSSage Weil 	spinlock_t i_unsafe_lock;
305de57606cSSage Weil 
306de57606cSSage Weil 	struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
307de57606cSSage Weil 	int i_snap_realm_counter; /* snap realm (if caps) */
308de57606cSSage Weil 	struct list_head i_snap_realm_item;
309de57606cSSage Weil 	struct list_head i_snap_flush_item;
310de57606cSSage Weil 
311de57606cSSage Weil 	struct work_struct i_wb_work;  /* writeback work */
312de57606cSSage Weil 	struct work_struct i_pg_inv_work;  /* page invalidation work */
313de57606cSSage Weil 
314de57606cSSage Weil 	struct work_struct i_vmtruncate_work;
315de57606cSSage Weil 
316de57606cSSage Weil 	struct inode vfs_inode; /* at end */
317de57606cSSage Weil };
318de57606cSSage Weil 
319de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
320de57606cSSage Weil {
321fbbccec9SNoah Watkins 	return container_of(inode, struct ceph_inode_info, vfs_inode);
322de57606cSSage Weil }
323de57606cSSage Weil 
324ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode)
325ad1fee96SYehuda Sadeh {
326ad1fee96SYehuda Sadeh 	return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
327ad1fee96SYehuda Sadeh }
328ad1fee96SYehuda Sadeh 
329ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb)
330ad1fee96SYehuda Sadeh {
331ad1fee96SYehuda Sadeh 	return (struct ceph_fs_client *)sb->s_fs_info;
332ad1fee96SYehuda Sadeh }
333ad1fee96SYehuda Sadeh 
3343d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode)
3353d14c5d2SYehuda Sadeh {
3363d14c5d2SYehuda Sadeh 	return ceph_inode(inode)->i_vino;
3373d14c5d2SYehuda Sadeh }
3383d14c5d2SYehuda Sadeh 
3393d14c5d2SYehuda Sadeh /*
3403d14c5d2SYehuda Sadeh  * ino_t is <64 bits on many architectures, blech.
3413d14c5d2SYehuda Sadeh  *
342ad1fee96SYehuda Sadeh  *               i_ino (kernel inode)   st_ino (userspace)
343ad1fee96SYehuda Sadeh  * i386          32                     32
344ad1fee96SYehuda Sadeh  * x86_64+ino32  64                     32
345ad1fee96SYehuda Sadeh  * x86_64        64                     64
346ad1fee96SYehuda Sadeh  */
347ad1fee96SYehuda Sadeh static inline u32 ceph_ino_to_ino32(ino_t ino)
348ad1fee96SYehuda Sadeh {
349ad1fee96SYehuda Sadeh 	ino ^= ino >> (sizeof(ino) * 8 - 32);
350ad1fee96SYehuda Sadeh 	if (!ino)
351ad1fee96SYehuda Sadeh 		ino = 1;
352ad1fee96SYehuda Sadeh 	return ino;
353ad1fee96SYehuda Sadeh }
354ad1fee96SYehuda Sadeh 
355ad1fee96SYehuda Sadeh /*
356ad1fee96SYehuda Sadeh  * kernel i_ino value
3573d14c5d2SYehuda Sadeh  */
3583d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
3593d14c5d2SYehuda Sadeh {
3603d14c5d2SYehuda Sadeh 	ino_t ino = (ino_t)vino.ino;  /* ^ (vino.snap << 20); */
3613d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32
362ad1fee96SYehuda Sadeh 	ino = ceph_ino_to_ino32(ino);
3633d14c5d2SYehuda Sadeh #endif
3643d14c5d2SYehuda Sadeh 	return ino;
3653d14c5d2SYehuda Sadeh }
3663d14c5d2SYehuda Sadeh 
367ad1fee96SYehuda Sadeh /*
368ad1fee96SYehuda Sadeh  * user-visible ino (stat, filldir)
369ad1fee96SYehuda Sadeh  */
370ad1fee96SYehuda Sadeh #if BITS_PER_LONG == 32
371ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
372ad1fee96SYehuda Sadeh {
373ad1fee96SYehuda Sadeh 	return ino;
374ad1fee96SYehuda Sadeh }
375ad1fee96SYehuda Sadeh #else
376ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
377ad1fee96SYehuda Sadeh {
378ad1fee96SYehuda Sadeh 	if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))
379ad1fee96SYehuda Sadeh 		ino = ceph_ino_to_ino32(ino);
380ad1fee96SYehuda Sadeh 	return ino;
381ad1fee96SYehuda Sadeh }
382ad1fee96SYehuda Sadeh #endif
383ad1fee96SYehuda Sadeh 
384ad1fee96SYehuda Sadeh 
3853d14c5d2SYehuda Sadeh /* for printf-style formatting */
3863d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
3873d14c5d2SYehuda Sadeh 
3883d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode)
3893d14c5d2SYehuda Sadeh {
3903d14c5d2SYehuda Sadeh 	return ceph_inode(inode)->i_vino.ino;
3913d14c5d2SYehuda Sadeh }
3923d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode)
3933d14c5d2SYehuda Sadeh {
3943d14c5d2SYehuda Sadeh 	return ceph_inode(inode)->i_vino.snap;
3953d14c5d2SYehuda Sadeh }
3963d14c5d2SYehuda Sadeh 
3973d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data)
3983d14c5d2SYehuda Sadeh {
3993d14c5d2SYehuda Sadeh 	struct ceph_vino *pvino = (struct ceph_vino *)data;
4003d14c5d2SYehuda Sadeh 	struct ceph_inode_info *ci = ceph_inode(inode);
4013d14c5d2SYehuda Sadeh 	return ci->i_vino.ino == pvino->ino &&
4023d14c5d2SYehuda Sadeh 		ci->i_vino.snap == pvino->snap;
4033d14c5d2SYehuda Sadeh }
4043d14c5d2SYehuda Sadeh 
4053d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb,
4063d14c5d2SYehuda Sadeh 					    struct ceph_vino vino)
4073d14c5d2SYehuda Sadeh {
4083d14c5d2SYehuda Sadeh 	ino_t t = ceph_vino_to_ino(vino);
4093d14c5d2SYehuda Sadeh 	return ilookup5(sb, t, ceph_ino_compare, &vino);
4103d14c5d2SYehuda Sadeh }
4113d14c5d2SYehuda Sadeh 
4123d14c5d2SYehuda Sadeh 
4133d14c5d2SYehuda Sadeh /*
4143d14c5d2SYehuda Sadeh  * Ceph inode.
4153d14c5d2SYehuda Sadeh  */
4163d14c5d2SYehuda Sadeh #define CEPH_I_COMPLETE  1  /* we have complete directory cached */
4173d14c5d2SYehuda Sadeh #define CEPH_I_NODELAY   4  /* do not delay cap release */
4183d14c5d2SYehuda Sadeh #define CEPH_I_FLUSH     8  /* do not delay flush of dirty metadata */
4193d14c5d2SYehuda Sadeh #define CEPH_I_NOFLUSH  16  /* do not flush dirty caps */
4203d14c5d2SYehuda Sadeh 
421de57606cSSage Weil static inline void ceph_i_clear(struct inode *inode, unsigned mask)
422de57606cSSage Weil {
423de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
424de57606cSSage Weil 
425de57606cSSage Weil 	spin_lock(&inode->i_lock);
426de57606cSSage Weil 	ci->i_ceph_flags &= ~mask;
427de57606cSSage Weil 	spin_unlock(&inode->i_lock);
428de57606cSSage Weil }
429de57606cSSage Weil 
430de57606cSSage Weil static inline void ceph_i_set(struct inode *inode, unsigned mask)
431de57606cSSage Weil {
432de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
433de57606cSSage Weil 
434de57606cSSage Weil 	spin_lock(&inode->i_lock);
435de57606cSSage Weil 	ci->i_ceph_flags |= mask;
436de57606cSSage Weil 	spin_unlock(&inode->i_lock);
437de57606cSSage Weil }
438de57606cSSage Weil 
439de57606cSSage Weil static inline bool ceph_i_test(struct inode *inode, unsigned mask)
440de57606cSSage Weil {
441de57606cSSage Weil 	struct ceph_inode_info *ci = ceph_inode(inode);
442de57606cSSage Weil 	bool r;
443de57606cSSage Weil 
444efa4c120SSage Weil 	spin_lock(&inode->i_lock);
445de57606cSSage Weil 	r = (ci->i_ceph_flags & mask) == mask;
446efa4c120SSage Weil 	spin_unlock(&inode->i_lock);
447de57606cSSage Weil 	return r;
448de57606cSSage Weil }
449de57606cSSage Weil 
450de57606cSSage Weil 
451de57606cSSage Weil /* find a specific frag @f */
452de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
453de57606cSSage Weil 						u32 f);
454de57606cSSage Weil 
455de57606cSSage Weil /*
456de57606cSSage Weil  * choose fragment for value @v.  copy frag content to pfrag, if leaf
457de57606cSSage Weil  * exists
458de57606cSSage Weil  */
459de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
460de57606cSSage Weil 			    struct ceph_inode_frag *pfrag,
461de57606cSSage Weil 			    int *found);
462de57606cSSage Weil 
463de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
464de57606cSSage Weil {
465de57606cSSage Weil 	return (struct ceph_dentry_info *)dentry->d_fsdata;
466de57606cSSage Weil }
467de57606cSSage Weil 
468de57606cSSage Weil static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
469de57606cSSage Weil {
470de57606cSSage Weil 	return ((loff_t)frag << 32) | (loff_t)off;
471de57606cSSage Weil }
472de57606cSSage Weil 
473de57606cSSage Weil /*
474de57606cSSage Weil  * caps helpers
475de57606cSSage Weil  */
476de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
477de57606cSSage Weil {
478de57606cSSage Weil 	return !RB_EMPTY_ROOT(&ci->i_caps);
479de57606cSSage Weil }
480de57606cSSage Weil 
481de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
482de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
483de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
484de57606cSSage Weil 				    struct ceph_cap *cap);
485de57606cSSage Weil 
486de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci)
487de57606cSSage Weil {
488de57606cSSage Weil 	int issued;
489de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
490de57606cSSage Weil 	issued = __ceph_caps_issued(ci, NULL);
491de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
492de57606cSSage Weil 	return issued;
493de57606cSSage Weil }
494de57606cSSage Weil 
495de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
496de57606cSSage Weil 					int touch)
497de57606cSSage Weil {
498de57606cSSage Weil 	int r;
499de57606cSSage Weil 	spin_lock(&ci->vfs_inode.i_lock);
500de57606cSSage Weil 	r = __ceph_caps_issued_mask(ci, mask, touch);
501de57606cSSage Weil 	spin_unlock(&ci->vfs_inode.i_lock);
502de57606cSSage Weil 	return r;
503de57606cSSage Weil }
504de57606cSSage Weil 
505de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
506de57606cSSage Weil {
507de57606cSSage Weil 	return ci->i_dirty_caps | ci->i_flushing_caps;
508de57606cSSage Weil }
509fca65b4aSSage Weil extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask);
510de57606cSSage Weil 
511de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
512de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci);
513de57606cSSage Weil 
514de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
515de57606cSSage Weil 
516de57606cSSage Weil /*
517de57606cSSage Weil  * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
518de57606cSSage Weil  */
519de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
520de57606cSSage Weil {
521de57606cSSage Weil 	int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
522de57606cSSage Weil 	if (w & CEPH_CAP_FILE_BUFFER)
523de57606cSSage Weil 		w |= CEPH_CAP_FILE_EXCL;  /* we want EXCL if dirty data */
524de57606cSSage Weil 	return w;
525de57606cSSage Weil }
526de57606cSSage Weil 
527de57606cSSage Weil /* what the mds thinks we want */
528de57606cSSage Weil extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
529de57606cSSage Weil 
53037151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc);
53137151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
53237151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
53337151668SYehuda Sadeh extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
53437151668SYehuda Sadeh 			     struct ceph_cap_reservation *ctx, int need);
53537151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
53637151668SYehuda Sadeh 			       struct ceph_cap_reservation *ctx);
5373d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client,
538de57606cSSage Weil 				    int *total, int *avail, int *used,
53985ccce43SSage Weil 				    int *reserved, int *min);
540de57606cSSage Weil 
541de57606cSSage Weil 
542de57606cSSage Weil 
543de57606cSSage Weil /*
544de57606cSSage Weil  * we keep buffered readdir results attached to file->private_data
545de57606cSSage Weil  */
5464918b6d1SSage Weil #define CEPH_F_SYNC     1
5479cfa1098SSage Weil #define CEPH_F_ATEND    2
5484918b6d1SSage Weil 
549de57606cSSage Weil struct ceph_file_info {
550252c6728SSage Weil 	short fmode;     /* initialized on open */
551252c6728SSage Weil 	short flags;     /* CEPH_F_* */
552de57606cSSage Weil 
553de57606cSSage Weil 	/* readdir: position within the dir */
554de57606cSSage Weil 	u32 frag;
555de57606cSSage Weil 	struct ceph_mds_request *last_readdir;
556de57606cSSage Weil 
557de57606cSSage Weil 	/* readdir: position within a frag */
558de57606cSSage Weil 	unsigned offset;       /* offset of last chunk, adjusted for . and .. */
559de57606cSSage Weil 	u64 next_offset;       /* offset of next chunk (last_name's + 1) */
560de57606cSSage Weil 	char *last_name;       /* last entry in previous chunk */
561de57606cSSage Weil 	struct dentry *dentry; /* next dentry (for dcache readdir) */
562de57606cSSage Weil 	unsigned long dir_release_count;
563de57606cSSage Weil 
564de57606cSSage Weil 	/* used for -o dirstat read() on directory thing */
565de57606cSSage Weil 	char *dir_info;
566de57606cSSage Weil 	int dir_info_len;
567de57606cSSage Weil };
568de57606cSSage Weil 
569de57606cSSage Weil 
570de57606cSSage Weil 
571de57606cSSage Weil /*
572de57606cSSage Weil  * A "snap realm" describes a subset of the file hierarchy sharing
573de57606cSSage Weil  * the same set of snapshots that apply to it.  The realms themselves
574de57606cSSage Weil  * are organized into a hierarchy, such that children inherit (some of)
575de57606cSSage Weil  * the snapshots of their parents.
576de57606cSSage Weil  *
577de57606cSSage Weil  * All inodes within the realm that have capabilities are linked into a
578de57606cSSage Weil  * per-realm list.
579de57606cSSage Weil  */
580de57606cSSage Weil struct ceph_snap_realm {
581de57606cSSage Weil 	u64 ino;
582de57606cSSage Weil 	atomic_t nref;
583a105f00cSSage Weil 	struct rb_node node;
584a105f00cSSage Weil 
585de57606cSSage Weil 	u64 created, seq;
586de57606cSSage Weil 	u64 parent_ino;
587de57606cSSage Weil 	u64 parent_since;   /* snapid when our current parent became so */
588de57606cSSage Weil 
589de57606cSSage Weil 	u64 *prior_parent_snaps;      /* snaps inherited from any parents we */
590de57606cSSage Weil 	int num_prior_parent_snaps;   /*  had prior to parent_since */
591de57606cSSage Weil 	u64 *snaps;                   /* snaps specific to this realm */
592de57606cSSage Weil 	int num_snaps;
593de57606cSSage Weil 
594de57606cSSage Weil 	struct ceph_snap_realm *parent;
595de57606cSSage Weil 	struct list_head children;       /* list of child realms */
596de57606cSSage Weil 	struct list_head child_item;
597de57606cSSage Weil 
598de57606cSSage Weil 	struct list_head empty_item;     /* if i have ref==0 */
599de57606cSSage Weil 
600ae00d4f3SSage Weil 	struct list_head dirty_item;     /* if realm needs new context */
601ae00d4f3SSage Weil 
602de57606cSSage Weil 	/* the current set of snaps for this realm */
603de57606cSSage Weil 	struct ceph_snap_context *cached_context;
604de57606cSSage Weil 
605de57606cSSage Weil 	struct list_head inodes_with_caps;
606de57606cSSage Weil 	spinlock_t inodes_with_caps_lock;
607de57606cSSage Weil };
608de57606cSSage Weil 
6093d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void)
6103d14c5d2SYehuda Sadeh {
6113d14c5d2SYehuda Sadeh 	int congestion_kb;
612de57606cSSage Weil 
613de57606cSSage Weil 	/*
6143d14c5d2SYehuda Sadeh 	 * Copied from NFS
6153d14c5d2SYehuda Sadeh 	 *
6163d14c5d2SYehuda Sadeh 	 * congestion size, scale with available memory.
6173d14c5d2SYehuda Sadeh 	 *
6183d14c5d2SYehuda Sadeh 	 *  64MB:    8192k
6193d14c5d2SYehuda Sadeh 	 * 128MB:   11585k
6203d14c5d2SYehuda Sadeh 	 * 256MB:   16384k
6213d14c5d2SYehuda Sadeh 	 * 512MB:   23170k
6223d14c5d2SYehuda Sadeh 	 *   1GB:   32768k
6233d14c5d2SYehuda Sadeh 	 *   2GB:   46340k
6243d14c5d2SYehuda Sadeh 	 *   4GB:   65536k
6253d14c5d2SYehuda Sadeh 	 *   8GB:   92681k
6263d14c5d2SYehuda Sadeh 	 *  16GB:  131072k
6273d14c5d2SYehuda Sadeh 	 *
6283d14c5d2SYehuda Sadeh 	 * This allows larger machines to have larger/more transfers.
6293d14c5d2SYehuda Sadeh 	 * Limit the default to 256M
630de57606cSSage Weil 	 */
6313d14c5d2SYehuda Sadeh 	congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
6323d14c5d2SYehuda Sadeh 	if (congestion_kb > 256*1024)
6333d14c5d2SYehuda Sadeh 		congestion_kb = 256*1024;
6343d14c5d2SYehuda Sadeh 
6353d14c5d2SYehuda Sadeh 	return congestion_kb;
636de57606cSSage Weil }
637de57606cSSage Weil 
638de57606cSSage Weil 
639de57606cSSage Weil 
640de57606cSSage Weil /* snap.c */
641de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
642de57606cSSage Weil 					       u64 ino);
643de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
644de57606cSSage Weil 				struct ceph_snap_realm *realm);
645de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
646de57606cSSage Weil 				struct ceph_snap_realm *realm);
647de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m,
648de57606cSSage Weil 				  void *p, void *e, bool deletion);
649de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
6502600d2ddSSage Weil 			     struct ceph_mds_session *session,
651de57606cSSage Weil 			     struct ceph_msg *msg);
652fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
653de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
654de57606cSSage Weil 				  struct ceph_cap_snap *capsnap);
655de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
656de57606cSSage Weil 
657de57606cSSage Weil /*
658de57606cSSage Weil  * a cap_snap is "pending" if it is still awaiting an in-progress
659de57606cSSage Weil  * sync write (that may/may not still update size, mtime, etc.).
660de57606cSSage Weil  */
661de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
662de57606cSSage Weil {
663de57606cSSage Weil 	return !list_empty(&ci->i_cap_snaps) &&
664de57606cSSage Weil 		list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap,
665de57606cSSage Weil 			   ci_item)->writing;
666de57606cSSage Weil }
667de57606cSSage Weil 
668de57606cSSage Weil /* inode.c */
669de57606cSSage Weil extern const struct inode_operations ceph_file_iops;
670de57606cSSage Weil 
671de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb);
672de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode);
673de57606cSSage Weil 
674de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb,
675de57606cSSage Weil 				    struct ceph_vino vino);
676de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent);
677de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued,
678de57606cSSage Weil 			       u32 truncate_seq, u64 truncate_size, u64 size);
679de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued,
680de57606cSSage Weil 				u64 time_warp_seq, struct timespec *ctime,
681de57606cSSage Weil 				struct timespec *mtime, struct timespec *atime);
682de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb,
683de57606cSSage Weil 			   struct ceph_mds_request *req,
684de57606cSSage Weil 			   struct ceph_mds_session *session);
685de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
686de57606cSSage Weil 				    struct ceph_mds_session *session);
687de57606cSSage Weil 
688de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask);
689de57606cSSage Weil 
690de57606cSSage Weil extern int ceph_inode_set_size(struct inode *inode, loff_t size);
691de57606cSSage Weil extern void __ceph_do_pending_vmtruncate(struct inode *inode);
6923c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode);
6933c6f6b79SSage Weil 
6943c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode);
6953c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode);
696de57606cSSage Weil 
697de57606cSSage Weil extern int ceph_do_getattr(struct inode *inode, int mask);
698b74c79e9SNick Piggin extern int ceph_permission(struct inode *inode, int mask, unsigned int flags);
699de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
700de57606cSSage Weil extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry,
701de57606cSSage Weil 			struct kstat *stat);
702de57606cSSage Weil 
703de57606cSSage Weil /* xattr.c */
704de57606cSSage Weil extern int ceph_setxattr(struct dentry *, const char *, const void *,
705de57606cSSage Weil 			 size_t, int);
706de57606cSSage Weil extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t);
707de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
708de57606cSSage Weil extern int ceph_removexattr(struct dentry *, const char *);
709de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
710de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
711de57606cSSage Weil 
712de57606cSSage Weil /* caps.c */
713de57606cSSage Weil extern const char *ceph_cap_string(int c);
714de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session,
715de57606cSSage Weil 			     struct ceph_msg *msg);
716de57606cSSage Weil extern int ceph_add_cap(struct inode *inode,
717de57606cSSage Weil 			struct ceph_mds_session *session, u64 cap_id,
718de57606cSSage Weil 			int fmode, unsigned issued, unsigned wanted,
719de57606cSSage Weil 			unsigned cap, unsigned seq, u64 realmino, int flags,
720de57606cSSage Weil 			struct ceph_cap_reservation *caps_reservation);
7217c1332b8SSage Weil extern void __ceph_remove_cap(struct ceph_cap *cap);
722de57606cSSage Weil static inline void ceph_remove_cap(struct ceph_cap *cap)
723de57606cSSage Weil {
724de57606cSSage Weil 	struct inode *inode = &cap->ci->vfs_inode;
725de57606cSSage Weil 	spin_lock(&inode->i_lock);
7267c1332b8SSage Weil 	__ceph_remove_cap(cap);
727de57606cSSage Weil 	spin_unlock(&inode->i_lock);
728de57606cSSage Weil }
72937151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc,
73037151668SYehuda Sadeh 			 struct ceph_cap *cap);
731de57606cSSage Weil 
732de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode);
733f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
7347ea80859SChristoph Hellwig extern int ceph_fsync(struct file *file, int datasync);
735de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
736de57606cSSage Weil 				    struct ceph_mds_session *session);
7372bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
7382bc50259SGreg Farnum 					     int mds);
739de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode);
740de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
741de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
742de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
743de57606cSSage Weil 				       struct ceph_snap_context *snapc);
744de57606cSSage Weil extern void __ceph_flush_snaps(struct ceph_inode_info *ci,
745e835124cSSage Weil 			       struct ceph_mds_session **psession,
746e835124cSSage Weil 			       int again);
747de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
748de57606cSSage Weil 			    struct ceph_mds_session *session);
749afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
750afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
751de57606cSSage Weil 
752de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode,
753de57606cSSage Weil 				     int mds, int drop, int unless, int force);
754de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
755de57606cSSage Weil 				      int mds, int drop, int unless);
756de57606cSSage Weil 
757de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
758de57606cSSage Weil 			 int *got, loff_t endoff);
759de57606cSSage Weil 
760de57606cSSage Weil /* for counting open files by mode */
761de57606cSSage Weil static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode)
762de57606cSSage Weil {
763de57606cSSage Weil 	ci->i_nr_by_mode[mode]++;
764de57606cSSage Weil }
765de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
766de57606cSSage Weil 
767de57606cSSage Weil /* addr.c */
768de57606cSSage Weil extern const struct address_space_operations ceph_aops;
769de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
770de57606cSSage Weil 
771de57606cSSage Weil /* file.c */
772de57606cSSage Weil extern const struct file_operations ceph_file_fops;
773de57606cSSage Weil extern const struct address_space_operations ceph_aops;
7743d14c5d2SYehuda Sadeh extern int ceph_copy_to_page_vector(struct page **pages,
7753d14c5d2SYehuda Sadeh 				    const char *data,
7763d14c5d2SYehuda Sadeh 				    loff_t off, size_t len);
7773d14c5d2SYehuda Sadeh extern int ceph_copy_from_page_vector(struct page **pages,
7783d14c5d2SYehuda Sadeh 				    char *data,
7793d14c5d2SYehuda Sadeh 				    loff_t off, size_t len);
7803d14c5d2SYehuda Sadeh extern struct page **ceph_alloc_page_vector(int num_pages, gfp_t flags);
781de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file);
782de57606cSSage Weil extern struct dentry *ceph_lookup_open(struct inode *dir, struct dentry *dentry,
783de57606cSSage Weil 				       struct nameidata *nd, int mode,
784de57606cSSage Weil 				       int locked_dir);
785de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp);
786de57606cSSage Weil 
787de57606cSSage Weil /* dir.c */
788de57606cSSage Weil extern const struct file_operations ceph_dir_fops;
789de57606cSSage Weil extern const struct inode_operations ceph_dir_iops;
79052dfb8acSSage Weil extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops,
791de57606cSSage Weil 	ceph_snapdir_dentry_ops;
792de57606cSSage Weil 
793de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
794468640e3SSage Weil extern int ceph_handle_snapdir(struct ceph_mds_request *req,
795468640e3SSage Weil 			       struct dentry *dentry, int err);
796de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
797de57606cSSage Weil 					 struct dentry *dentry, int err);
798de57606cSSage Weil 
799de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn);
800de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn);
801de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn);
80281a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
8036c0f3af7SSage Weil extern unsigned ceph_dentry_hash(struct dentry *dn);
804*5f21c96dSSage Weil extern struct inode *ceph_get_dentry_parent_inode(struct dentry *dentry);
805de57606cSSage Weil 
806de57606cSSage Weil /*
807de57606cSSage Weil  * our d_ops vary depending on whether the inode is live,
808de57606cSSage Weil  * snapshotted (read-only), or a virtual ".snap" directory.
809de57606cSSage Weil  */
810de57606cSSage Weil int ceph_init_dentry(struct dentry *dentry);
811de57606cSSage Weil 
812de57606cSSage Weil 
813de57606cSSage Weil /* ioctl.c */
814de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
815de57606cSSage Weil 
816de57606cSSage Weil /* export.c */
817de57606cSSage Weil extern const struct export_operations ceph_export_ops;
818de57606cSSage Weil 
81940819f6fSGreg Farnum /* locks.c */
82040819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
82140819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
82240819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
82340819f6fSGreg Farnum extern int ceph_encode_locks(struct inode *i, struct ceph_pagelist *p,
82440819f6fSGreg Farnum 			     int p_locks, int f_locks);
82540819f6fSGreg Farnum extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c);
82640819f6fSGreg Farnum 
8273d14c5d2SYehuda Sadeh /* debugfs.c */
8283d14c5d2SYehuda Sadeh extern int ceph_fs_debugfs_init(struct ceph_fs_client *client);
8293d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
8303d14c5d2SYehuda Sadeh 
831de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */
832