[linux-block.git] / fs / ceph / mds_client.h

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef _FS_CEPH_MDS_CLIENT_H
#define _FS_CEPH_MDS_CLIENT_H

#include <linux/completion.h>
#include <linux/kref.h>
#include <linux/list.h>
#include <linux/mutex.h>
#include <linux/rbtree.h>
#include <linux/spinlock.h>
#include <linux/refcount.h>
#include <linux/utsname.h>

#include <linux/ceph/types.h>
#include <linux/ceph/messenger.h>
#include <linux/ceph/mdsmap.h>
#include <linux/ceph/auth.h>

/* The first 8 bits are reserved for old ceph releases */
enum ceph_feature_type {
	CEPHFS_FEATURE_MIMIC = 8,
	CEPHFS_FEATURE_REPLY_ENCODING,
	CEPHFS_FEATURE_RECLAIM_CLIENT,
	CEPHFS_FEATURE_LAZY_CAP_WANTED,
	CEPHFS_FEATURE_MULTI_RECONNECT,

	CEPHFS_FEATURE_MAX = CEPHFS_FEATURE_MULTI_RECONNECT,
};

/*
 * This will always have the highest feature bit value
 * as the last element of the array.
 */
#define CEPHFS_FEATURES_CLIENT_SUPPORTED {	\
	0, 1, 2, 3, 4, 5, 6, 7,			\
	CEPHFS_FEATURE_MIMIC,			\
	CEPHFS_FEATURE_REPLY_ENCODING,		\
	CEPHFS_FEATURE_LAZY_CAP_WANTED,		\
	CEPHFS_FEATURE_MULTI_RECONNECT,		\
						\
	CEPHFS_FEATURE_MAX,			\
}
#define CEPHFS_FEATURES_CLIENT_REQUIRED {}

/*
 * Some lock dependencies:
 *
 * session->s_mutex
 *         mdsc->mutex
 *
 *         mdsc->snap_rwsem
 *
 *         ci->i_ceph_lock
 *                 mdsc->snap_flush_lock
 *                 mdsc->cap_delay_lock
 *
 */

struct ceph_fs_client;
struct ceph_cap;

/*
 * parsed info about a single inode.  pointers are into the encoded
 * on-wire structures within the mds reply message payload.
 */
struct ceph_mds_reply_info_in {
	struct ceph_mds_reply_inode *in;
	struct ceph_dir_layout dir_layout;
	u32 symlink_len;
	char *symlink;
	u32 xattr_len;
	char *xattr_data;
	u64 inline_version;
	u32 inline_len;
	char *inline_data;
	u32 pool_ns_len;
	char *pool_ns_data;
	u64 max_bytes;
	u64 max_files;
	s32 dir_pin;
	struct ceph_timespec btime;
	struct ceph_timespec snap_btime;
	u64 change_attr;
};

struct ceph_mds_reply_dir_entry {
	char                          *name;
	u32                           name_len;
	struct ceph_mds_reply_lease   *lease;
	struct ceph_mds_reply_info_in inode;
	loff_t			      offset;
};

/*
 * parsed info about an mds reply, including information about
 * either: 1) the target inode and/or its parent directory and dentry,
 * and directory contents (for readdir results), or
 * 2) the file range lock info (for fcntl F_GETLK results).
 */
struct ceph_mds_reply_info_parsed {
	struct ceph_mds_reply_head    *head;

	/* trace */
	struct ceph_mds_reply_info_in diri, targeti;
	struct ceph_mds_reply_dirfrag *dirfrag;
	char                          *dname;
	u32                           dname_len;
	struct ceph_mds_reply_lease   *dlease;

	/* extra */
	union {
		/* for fcntl F_GETLK results */
		struct ceph_filelock *filelock_reply;

		/* for readdir results */
		struct {
			struct ceph_mds_reply_dirfrag *dir_dir;
			size_t			      dir_buf_size;
			int                           dir_nr;
			bool			      dir_end;
			bool			      dir_complete;
			bool			      hash_order;
			bool			      offset_hash;
			struct ceph_mds_reply_dir_entry  *dir_entries;
		};

		/* for create results */
		struct {
			bool has_create_ino;
			u64 ino;
		};
	};

	/* encoded blob describing snapshot contexts for certain
	   operations (e.g., open) */
	void *snapblob;
	int snapblob_len;
};


/*
 * cap releases are batched and sent to the MDS en masse.
 *
 * Account for per-message overhead of mds_cap_release header
 * and __le32 for osd epoch barrier trailing field.
 */
#define CEPH_CAPS_PER_RELEASE ((PAGE_SIZE - sizeof(u32) -		\
				sizeof(struct ceph_mds_cap_release)) /	\
			        sizeof(struct ceph_mds_cap_item))


/*
 * state associated with each MDS<->client session
 */
enum {
	CEPH_MDS_SESSION_NEW = 1,
	CEPH_MDS_SESSION_OPENING = 2,
	CEPH_MDS_SESSION_OPEN = 3,
	CEPH_MDS_SESSION_HUNG = 4,
	CEPH_MDS_SESSION_RESTARTING = 5,
	CEPH_MDS_SESSION_RECONNECTING = 6,
	CEPH_MDS_SESSION_CLOSING = 7,
	CEPH_MDS_SESSION_CLOSED = 8,
	CEPH_MDS_SESSION_REJECTED = 9,
};

struct ceph_mds_session {
	struct ceph_mds_client *s_mdsc;
	int               s_mds;
	int               s_state;
	unsigned long     s_ttl;      /* time until mds kills us */
	unsigned long	  s_features;
	u64               s_seq;      /* incoming msg seq # */
	struct mutex      s_mutex;    /* serialize session messages */

	struct ceph_connection s_con;

	struct ceph_auth_handshake s_auth;

	/* protected by s_gen_ttl_lock */
	spinlock_t        s_gen_ttl_lock;
	u32               s_cap_gen;  /* inc each time we get mds stale msg */
	unsigned long     s_cap_ttl;  /* when session caps expire */

	/* protected by s_cap_lock */
	spinlock_t        s_cap_lock;
	refcount_t        s_ref;
	struct list_head  s_caps;     /* all caps issued by this session */
	struct ceph_cap  *s_cap_iterator;
	int               s_nr_caps;
	int               s_num_cap_releases;
	int		  s_cap_reconnect;
	int		  s_readonly;
	struct list_head  s_cap_releases; /* waiting cap_release messages */
	struct work_struct s_cap_release_work;

	/* protected by mutex */
	struct list_head  s_cap_flushing;     /* inodes w/ flushing caps */
	unsigned long     s_renew_requested; /* last time we sent a renew req */
	u64               s_renew_seq;

	struct list_head  s_waiting;  /* waiting requests */
	struct list_head  s_unsafe;   /* unsafe requests */
};

/*
 * modes of choosing which MDS to send a request to
 */
enum {
	USE_ANY_MDS,
	USE_RANDOM_MDS,
	USE_AUTH_MDS,   /* prefer authoritative mds for this metadata item */
};

struct ceph_mds_request;
struct ceph_mds_client;

/*
 * request completion callback
 */
typedef void (*ceph_mds_request_callback_t) (struct ceph_mds_client *mdsc,
					     struct ceph_mds_request *req);
/*
 * wait for request completion callback
 */
typedef int (*ceph_mds_request_wait_callback_t) (struct ceph_mds_client *mdsc,
						 struct ceph_mds_request *req);

/*
 * an in-flight mds request
 */
struct ceph_mds_request {
	u64 r_tid;                   /* transaction id */
	struct rb_node r_node;
	struct ceph_mds_client *r_mdsc;

	struct kref       r_kref;
	int r_op;                    /* mds op code */

	/* operation on what? */
	struct inode *r_inode;              /* arg1 */
	struct dentry *r_dentry;            /* arg1 */
	struct dentry *r_old_dentry;        /* arg2: rename from or link from */
	struct inode *r_old_dentry_dir;     /* arg2: old dentry's parent dir */
	char *r_path1, *r_path2;
	struct ceph_vino r_ino1, r_ino2;

	struct inode *r_parent;		    /* parent dir inode */
	struct inode *r_target_inode;       /* resulting inode */

#define CEPH_MDS_R_DIRECT_IS_HASH	(1) /* r_direct_hash is valid */
#define CEPH_MDS_R_ABORTED		(2) /* call was aborted */
#define CEPH_MDS_R_GOT_UNSAFE		(3) /* got an unsafe reply */
#define CEPH_MDS_R_GOT_SAFE		(4) /* got a safe reply */
#define CEPH_MDS_R_GOT_RESULT		(5) /* got a result */
#define CEPH_MDS_R_DID_PREPOPULATE	(6) /* prepopulated readdir */
#define CEPH_MDS_R_PARENT_LOCKED	(7) /* is r_parent->i_rwsem wlocked? */
#define CEPH_MDS_R_ASYNC		(8) /* async request */
	unsigned long	r_req_flags;

	struct mutex r_fill_mutex;

	union ceph_mds_request_args r_args;
	int r_fmode;        /* file mode, if expecting cap */
	kuid_t r_uid;
	kgid_t r_gid;
	int r_request_release_offset;
	struct timespec64 r_stamp;

	/* for choosing which mds to send this request to */
	int r_direct_mode;
	u32 r_direct_hash;      /* choose dir frag based on this dentry hash */

	/* data payload is used for xattr ops */
	struct ceph_pagelist *r_pagelist;

	/* what caps shall we drop? */
	int r_inode_drop, r_inode_unless;
	int r_dentry_drop, r_dentry_unless;
	int r_old_dentry_drop, r_old_dentry_unless;
	struct inode *r_old_inode;
	int r_old_inode_drop, r_old_inode_unless;

	struct ceph_msg  *r_request;  /* original request */
	struct ceph_msg  *r_reply;
	struct ceph_mds_reply_info_parsed r_reply_info;
	struct page *r_locked_page;
	int r_err;
	int r_num_caps;
	u32               r_readdir_offset;

	unsigned long r_timeout;  /* optional.  jiffies, 0 is "wait forever" */
	unsigned long r_started;  /* start time to measure timeout against */
	unsigned long r_request_started; /* start time for mds request only,
					    used to measure lease durations */

	/* link unsafe requests to parent directory, for fsync */
	struct inode	*r_unsafe_dir;
	struct list_head r_unsafe_dir_item;

	/* unsafe requests that modify the target inode */
	struct list_head r_unsafe_target_item;

	struct ceph_mds_session *r_session;

	int               r_attempts;   /* resend attempts */
	int               r_num_fwd;    /* number of forward attempts */
	int               r_resend_mds; /* mds to resend to next, if any*/
	u32               r_sent_on_mseq; /* cap mseq request was sent at*/

	struct list_head  r_wait;
	struct completion r_completion;
	struct completion r_safe_completion;
	ceph_mds_request_callback_t r_callback;
	ceph_mds_request_wait_callback_t r_wait_for_completion;
	struct list_head  r_unsafe_item;  /* per-session unsafe list item */

	long long	  r_dir_release_cnt;
	long long	  r_dir_ordered_cnt;
	int		  r_readdir_cache_idx;

	struct ceph_cap_reservation r_caps_reservation;
};

struct ceph_pool_perm {
	struct rb_node node;
	int perm;
	s64 pool;
	size_t pool_ns_len;
	char pool_ns[];
};

struct ceph_snapid_map {
	struct rb_node node;
	struct list_head lru;
	atomic_t ref;
	u64 snap;
	dev_t dev;
	unsigned long last_used;
};

/*
 * node for list of quotarealm inodes that are not visible from the filesystem
 * mountpoint, but required to handle, e.g. quotas.
 */
struct ceph_quotarealm_inode {
	struct rb_node node;
	u64 ino;
	unsigned long timeout; /* last time a lookup failed for this inode */
	struct mutex mutex;
	struct inode *inode;
};

struct cap_wait {
	struct list_head	list;
	unsigned long		ino;
	pid_t			tgid;
	int			need;
	int			want;
};

/*
 * mds client state
 */
struct ceph_mds_client {
	struct ceph_fs_client  *fsc;
	struct mutex            mutex;         /* all nested structures */

	struct ceph_mdsmap      *mdsmap;
	struct completion       safe_umount_waiters;
	wait_queue_head_t       session_close_wq;
	struct list_head        waiting_for_map;
	int 			mdsmap_err;

	struct ceph_mds_session **sessions;    /* NULL for mds if no session */
	atomic_t		num_sessions;
	int                     max_sessions;  /* len of s_mds_sessions */
	int                     stopping;      /* true if shutting down */

	atomic64_t		quotarealms_count; /* # realms with quota */
	/*
	 * We keep a list of inodes we don't see in the mountpoint but that we
	 * need to track quota realms.
	 */
	struct rb_root		quotarealms_inodes;
	struct mutex		quotarealms_inodes_mutex;

	/*
	 * snap_rwsem will cover cap linkage into snaprealms, and
	 * realm snap contexts.  (later, we can do per-realm snap
	 * contexts locks..)  the empty list contains realms with no
	 * references (implying they contain no inodes with caps) that
	 * should be destroyed.
	 */
	u64			last_snap_seq;
	struct rw_semaphore     snap_rwsem;
	struct rb_root          snap_realms;
	struct list_head        snap_empty;
	int			num_snap_realms;
	spinlock_t              snap_empty_lock;  /* protect snap_empty */

	u64                    last_tid;      /* most recent mds request */
	u64                    oldest_tid;    /* oldest incomplete mds request,
						 excluding setfilelock requests */
	struct rb_root         request_tree;  /* pending mds requests */
	struct delayed_work    delayed_work;  /* delayed work */
	unsigned long    last_renew_caps;  /* last time we renewed our caps */
	struct list_head cap_delay_list;   /* caps with delayed release */
	spinlock_t       cap_delay_lock;   /* protects cap_delay_list */
	struct list_head snap_flush_list;  /* cap_snaps ready to flush */
	spinlock_t       snap_flush_lock;

	u64               last_cap_flush_tid;
	struct list_head  cap_flush_list;
	struct list_head  cap_dirty;        /* inodes with dirty caps */
	struct list_head  cap_dirty_migrating; /* ...that are migration... */
	int               num_cap_flushing; /* # caps we are flushing */
	spinlock_t        cap_dirty_lock;   /* protects above items */
	wait_queue_head_t cap_flushing_wq;

	struct work_struct cap_reclaim_work;
	atomic_t	   cap_reclaim_pending;

	/*
	 * Cap reservations
	 *
	 * Maintain a global pool of preallocated struct ceph_caps, referenced
	 * by struct ceph_caps_reservations.  This ensures that we preallocate
	 * memory needed to successfully process an MDS response.  (If an MDS
	 * sends us cap information and we fail to process it, we will have
	 * problems due to the client and MDS being out of sync.)
	 *
	 * Reservations are 'owned' by a ceph_cap_reservation context.
	 */
	spinlock_t	caps_list_lock;
	struct		list_head caps_list; /* unused (reserved or
						unreserved) */
	struct		list_head cap_wait_list;
	int		caps_total_count;    /* total caps allocated */
	int		caps_use_count;      /* in use */
	int		caps_use_max;	     /* max used caps */
	int		caps_reserve_count;  /* unused, reserved */
	int		caps_avail_count;    /* unused, unreserved */
	int		caps_min_count;      /* keep at least this many
						(unreserved) */
	spinlock_t	  dentry_list_lock;
	struct list_head  dentry_leases;     /* fifo list */
	struct list_head  dentry_dir_leases; /* lru list */

	spinlock_t		snapid_map_lock;
	struct rb_root		snapid_map_tree;
	struct list_head	snapid_map_lru;

	struct rw_semaphore     pool_perm_rwsem;
	struct rb_root		pool_perm_tree;

	char nodename[__NEW_UTS_LEN + 1];
};

extern const char *ceph_mds_op_name(int op);

extern struct ceph_mds_session *
__ceph_lookup_mds_session(struct ceph_mds_client *, int mds);

extern const char *ceph_session_state_name(int s);

extern struct ceph_mds_session *
ceph_get_mds_session(struct ceph_mds_session *s);
extern void ceph_put_mds_session(struct ceph_mds_session *s);

extern int ceph_send_msg_mds(struct ceph_mds_client *mdsc,
			     struct ceph_msg *msg, int mds);

extern int ceph_mdsc_init(struct ceph_fs_client *fsc);
extern void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc);
extern void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc);
extern void ceph_mdsc_destroy(struct ceph_fs_client *fsc);

extern void ceph_mdsc_sync(struct ceph_mds_client *mdsc);

extern void ceph_invalidate_dir_request(struct ceph_mds_request *req);
extern int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
					   struct inode *dir);
extern struct ceph_mds_request *
ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode);
extern int ceph_mdsc_submit_request(struct ceph_mds_client *mdsc,
				    struct inode *dir,
				    struct ceph_mds_request *req);
extern int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
				struct inode *dir,
				struct ceph_mds_request *req);
static inline void ceph_mdsc_get_request(struct ceph_mds_request *req)
{
	kref_get(&req->r_kref);
}
extern void ceph_mdsc_release_request(struct kref *kref);
static inline void ceph_mdsc_put_request(struct ceph_mds_request *req)
{
	kref_put(&req->r_kref, ceph_mdsc_release_request);
}

extern void __ceph_queue_cap_release(struct ceph_mds_session *session,
				    struct ceph_cap *cap);
extern void ceph_flush_cap_releases(struct ceph_mds_client *mdsc,
				    struct ceph_mds_session *session);
extern void ceph_queue_cap_reclaim_work(struct ceph_mds_client *mdsc);
extern void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr);
extern int ceph_iterate_session_caps(struct ceph_mds_session *session,
				     int (*cb)(struct inode *,
					       struct ceph_cap *, void *),
				     void *arg);
extern void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc);

static inline void ceph_mdsc_free_path(char *path, int len)
{
	if (path)
		__putname(path - (PATH_MAX - 1 - len));
}

extern char *ceph_mdsc_build_path(struct dentry *dentry, int *plen, u64 *base,
				  int stop_on_nosnap);

extern void __ceph_mdsc_drop_dentry_lease(struct dentry *dentry);
extern void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
				     struct dentry *dentry, char action,
				     u32 seq);

extern void ceph_mdsc_handle_mdsmap(struct ceph_mds_client *mdsc,
				    struct ceph_msg *msg);
extern void ceph_mdsc_handle_fsmap(struct ceph_mds_client *mdsc,
				   struct ceph_msg *msg);

extern struct ceph_mds_session *
ceph_mdsc_open_export_target_session(struct ceph_mds_client *mdsc, int target);
extern void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
					  struct ceph_mds_session *session);

extern int ceph_trim_caps(struct ceph_mds_client *mdsc,
			  struct ceph_mds_session *session,
			  int max_caps);
#endif
Commit	Line	Data
b2441318	1	/* SPDX-License-Identifier: GPL-2.0 */
2f2dc053 SW	2	#ifndef _FS_CEPH_MDS_CLIENT_H
	3	#define _FS_CEPH_MDS_CLIENT_H
	4
	5	#include <linux/completion.h>
153c8e6b	6	#include <linux/kref.h>
2f2dc053 SW	7	#include <linux/list.h>
2f2dc053 SW	8	#include <linux/mutex.h>
44ca18f2	9	#include <linux/rbtree.h>
2f2dc053	10	#include <linux/spinlock.h>
3997c01d	11	#include <linux/refcount.h>
717e6f28	12	#include <linux/utsname.h>
2f2dc053	13
3d14c5d2 YS	14	#include <linux/ceph/types.h>
	15	#include <linux/ceph/messenger.h>
	16	#include <linux/ceph/mdsmap.h>
6c4a1915	17	#include <linux/ceph/auth.h>
2f2dc053	18
342ce182	19	/* The first 8 bits are reserved for old ceph releases */
9ba1e224 XL	20	enum ceph_feature_type {
	21	CEPHFS_FEATURE_MIMIC = 8,
	22	CEPHFS_FEATURE_REPLY_ENCODING,
	23	CEPHFS_FEATURE_RECLAIM_CLIENT,
	24	CEPHFS_FEATURE_LAZY_CAP_WANTED,
	25	CEPHFS_FEATURE_MULTI_RECONNECT,
d2f8bb27	26
9ba1e224 XL	27	CEPHFS_FEATURE_MAX = CEPHFS_FEATURE_MULTI_RECONNECT,
	28	};
	29
	30	/*
	31	* This will always have the highest feature bit value
	32	* as the last element of the array.
	33	*/
	34	#define CEPHFS_FEATURES_CLIENT_SUPPORTED { \
d2f8bb27 YZ	35	0, 1, 2, 3, 4, 5, 6, 7, \
d2f8bb27 YZ	36	CEPHFS_FEATURE_MIMIC, \
b37fe1f9	37	CEPHFS_FEATURE_REPLY_ENCODING, \
d2f8bb27	38	CEPHFS_FEATURE_LAZY_CAP_WANTED, \
81c5a148	39	CEPHFS_FEATURE_MULTI_RECONNECT, \
9ba1e224 XL	40	\
9ba1e224 XL	41	CEPHFS_FEATURE_MAX, \
342ce182	42	}
342ce182 YZ	43	#define CEPHFS_FEATURES_CLIENT_REQUIRED {}
342ce182 YZ	44
2f2dc053 SW	45	/*
	46	* Some lock dependencies:
	47	*
	48	* session->s_mutex
	49	* mdsc->mutex
	50	*
	51	* mdsc->snap_rwsem
	52	*
be655596	53	* ci->i_ceph_lock
2f2dc053 SW	54	* mdsc->snap_flush_lock
	55	* mdsc->cap_delay_lock
	56	*
	57	*/
	58
3d14c5d2	59	struct ceph_fs_client;
2f2dc053 SW	60	struct ceph_cap;
	61
	62	/*
	63	* parsed info about a single inode. pointers are into the encoded
	64	* on-wire structures within the mds reply message payload.
	65	*/
	66	struct ceph_mds_reply_info_in {
	67	struct ceph_mds_reply_inode *in;
14303d20	68	struct ceph_dir_layout dir_layout;
2f2dc053 SW	69	u32 symlink_len;
	70	char *symlink;
	71	u32 xattr_len;
	72	char *xattr_data;
fb01d1f8 YZ	73	u64 inline_version;
	74	u32 inline_len;
	75	char *inline_data;
5ea5c5e0	76	u32 pool_ns_len;
779fe0fb	77	char *pool_ns_data;
fb18a575 LH	78	u64 max_bytes;
fb18a575 LH	79	u64 max_files;
08796873	80	s32 dir_pin;
245ce991	81	struct ceph_timespec btime;
193e7b37	82	struct ceph_timespec snap_btime;
a35ead31	83	u64 change_attr;
2f2dc053 SW	84	};
2f2dc053 SW	85
2a5beea3 YZ	86	struct ceph_mds_reply_dir_entry {
	87	char *name;
	88	u32 name_len;
	89	struct ceph_mds_reply_lease *lease;
	90	struct ceph_mds_reply_info_in inode;
8974eebd	91	loff_t offset;
2a5beea3 YZ	92	};
2a5beea3 YZ	93
2f2dc053	94	/*
25933abd HS	95	* parsed info about an mds reply, including information about
	96	* either: 1) the target inode and/or its parent directory and dentry,
	97	* and directory contents (for readdir results), or
	98	* 2) the file range lock info (for fcntl F_GETLK results).
2f2dc053 SW	99	*/
	100	struct ceph_mds_reply_info_parsed {
	101	struct ceph_mds_reply_head *head;
	102
25933abd	103	/* trace */
2f2dc053 SW	104	struct ceph_mds_reply_info_in diri, targeti;
	105	struct ceph_mds_reply_dirfrag *dirfrag;
	106	char *dname;
	107	u32 dname_len;
	108	struct ceph_mds_reply_lease *dlease;
	109
25933abd HS	110	/* extra */
	111	union {
	112	/* for fcntl F_GETLK results */
	113	struct ceph_filelock *filelock_reply;
	114
	115	/* for readdir results */
	116	struct {
	117	struct ceph_mds_reply_dirfrag *dir_dir;
54008399	118	size_t dir_buf_size;
25933abd	119	int dir_nr;
f3c4ebe6	120	bool dir_end;
79162547	121	bool dir_complete;
f3c4ebe6	122	bool hash_order;
79162547	123	bool offset_hash;
2a5beea3	124	struct ceph_mds_reply_dir_entry *dir_entries;
25933abd	125	};
6e8575fa SL	126
	127	/* for create results */
	128	struct {
	129	bool has_create_ino;
	130	u64 ino;
	131	};
25933abd	132	};
2f2dc053 SW	133
	134	/* encoded blob describing snapshot contexts for certain
	135	operations (e.g., open) */
	136	void *snapblob;
	137	int snapblob_len;
	138	};
	139
	140
	141	/*
	142	* cap releases are batched and sent to the MDS en masse.
92475f05 JL	143	*
	144	* Account for per-message overhead of mds_cap_release header
	145	* and __le32 for osd epoch barrier trailing field.
2f2dc053	146	*/
92475f05	147	#define CEPH_CAPS_PER_RELEASE ((PAGE_SIZE - sizeof(u32) - \
2f2dc053	148	sizeof(struct ceph_mds_cap_release)) / \
92475f05	149	sizeof(struct ceph_mds_cap_item))
2f2dc053 SW	150
	151
	152	/*
	153	* state associated with each MDS<->client session
	154	*/
	155	enum {
	156	CEPH_MDS_SESSION_NEW = 1,
	157	CEPH_MDS_SESSION_OPENING = 2,
	158	CEPH_MDS_SESSION_OPEN = 3,
	159	CEPH_MDS_SESSION_HUNG = 4,
7e6906c1 YZ	160	CEPH_MDS_SESSION_RESTARTING = 5,
	161	CEPH_MDS_SESSION_RECONNECTING = 6,
	162	CEPH_MDS_SESSION_CLOSING = 7,
4d681c2f XL	163	CEPH_MDS_SESSION_CLOSED = 8,
4d681c2f XL	164	CEPH_MDS_SESSION_REJECTED = 9,
2f2dc053 SW	165	};
	166
	167	struct ceph_mds_session {
	168	struct ceph_mds_client *s_mdsc;
	169	int s_mds;
	170	int s_state;
	171	unsigned long s_ttl; /* time until mds kills us */
84bf3950	172	unsigned long s_features;
2f2dc053 SW	173	u64 s_seq; /* incoming msg seq # */
	174	struct mutex s_mutex; /* serialize session messages */
	175
	176	struct ceph_connection s_con;
	177
6c4a1915	178	struct ceph_auth_handshake s_auth;
4e7a5dcd	179
d8fb02ab AE	180	/* protected by s_gen_ttl_lock */
d8fb02ab AE	181	spinlock_t s_gen_ttl_lock;
2f2dc053 SW	182	u32 s_cap_gen; /* inc each time we get mds stale msg */
2f2dc053 SW	183	unsigned long s_cap_ttl; /* when session caps expire */
d8fb02ab AE	184
	185	/* protected by s_cap_lock */
	186	spinlock_t s_cap_lock;
045100cd	187	refcount_t s_ref;
2f2dc053	188	struct list_head s_caps; /* all caps issued by this session */
e3ec8d68	189	struct ceph_cap *s_cap_iterator;
533a2818	190	int s_nr_caps;
2f2dc053	191	int s_num_cap_releases;
99a9c273	192	int s_cap_reconnect;
03f4fcb0	193	int s_readonly;
2f2dc053	194	struct list_head s_cap_releases; /* waiting cap_release messages */
e3ec8d68	195	struct work_struct s_cap_release_work;
2f2dc053 SW	196
	197	/* protected by mutex */
	198	struct list_head s_cap_flushing; /* inodes w/ flushing caps */
2f2dc053 SW	199	unsigned long s_renew_requested; /* last time we sent a renew req */
	200	u64 s_renew_seq;
	201
2f2dc053 SW	202	struct list_head s_waiting; /* waiting requests */
	203	struct list_head s_unsafe; /* unsafe requests */
	204	};
	205
	206	/*
	207	* modes of choosing which MDS to send a request to
	208	*/
	209	enum {
	210	USE_ANY_MDS,
	211	USE_RANDOM_MDS,
	212	USE_AUTH_MDS, /* prefer authoritative mds for this metadata item */
	213	};
	214
	215	struct ceph_mds_request;
	216	struct ceph_mds_client;
	217
	218	/*
	219	* request completion callback
	220	*/
	221	typedef void (ceph_mds_request_callback_t) (struct ceph_mds_client mdsc,
	222	struct ceph_mds_request *req);
9280be24 YZ	223	/*
	224	* wait for request completion callback
	225	*/
	226	typedef int (ceph_mds_request_wait_callback_t) (struct ceph_mds_client mdsc,
	227	struct ceph_mds_request *req);
2f2dc053 SW	228
	229	/*
	230	* an in-flight mds request
	231	*/
	232	struct ceph_mds_request {
	233	u64 r_tid; /* transaction id */
44ca18f2	234	struct rb_node r_node;
37151668	235	struct ceph_mds_client *r_mdsc;
2f2dc053	236
045100cd	237	struct kref r_kref;
2f2dc053	238	int r_op; /* mds op code */
2f2dc053 SW	239
	240	/* operation on what? */
	241	struct inode r_inode; / arg1 */
	242	struct dentry r_dentry; / arg1 */
	243	struct dentry r_old_dentry; / arg2: rename from or link from */
41b02e1f	244	struct inode r_old_dentry_dir; / arg2: old dentry's parent dir */
2f2dc053 SW	245	char r_path1, r_path2;
	246	struct ceph_vino r_ino1, r_ino2;
	247
3dd69aab	248	struct inode r_parent; / parent dir inode */
2f2dc053 SW	249	struct inode r_target_inode; / resulting inode */
2f2dc053 SW	250
bc2de10d JL	251	#define CEPH_MDS_R_DIRECT_IS_HASH (1) /* r_direct_hash is valid */
	252	#define CEPH_MDS_R_ABORTED (2) /* call was aborted */
	253	#define CEPH_MDS_R_GOT_UNSAFE (3) /* got an unsafe reply */
	254	#define CEPH_MDS_R_GOT_SAFE (4) /* got a safe reply */
	255	#define CEPH_MDS_R_GOT_RESULT (5) /* got a result */
	256	#define CEPH_MDS_R_DID_PREPOPULATE (6) /* prepopulated readdir */
3dd69aab	257	#define CEPH_MDS_R_PARENT_LOCKED (7) /* is r_parent->i_rwsem wlocked? */
3bb48b41	258	#define CEPH_MDS_R_ASYNC (8) /* async request */
bc2de10d JL	259	unsigned long r_req_flags;
bc2de10d JL	260
b4556396 SW	261	struct mutex r_fill_mutex;
b4556396 SW	262
2f2dc053 SW	263	union ceph_mds_request_args r_args;
2f2dc053 SW	264	int r_fmode; /* file mode, if expecting cap */
ff3d0046 EB	265	kuid_t r_uid;
ff3d0046 EB	266	kgid_t r_gid;
c36d6414	267	int r_request_release_offset;
0ed1e90a	268	struct timespec64 r_stamp;
2f2dc053 SW	269
	270	/* for choosing which mds to send this request to */
	271	int r_direct_mode;
	272	u32 r_direct_hash; /* choose dir frag based on this dentry hash */
2f2dc053 SW	273
2f2dc053 SW	274	/* data payload is used for xattr ops */
25e6bae3	275	struct ceph_pagelist *r_pagelist;
2f2dc053 SW	276
	277	/* what caps shall we drop? */
	278	int r_inode_drop, r_inode_unless;
	279	int r_dentry_drop, r_dentry_unless;
	280	int r_old_dentry_drop, r_old_dentry_unless;
	281	struct inode *r_old_inode;
	282	int r_old_inode_drop, r_old_inode_unless;
	283
	284	struct ceph_msg r_request; / original request */
	285	struct ceph_msg *r_reply;
	286	struct ceph_mds_reply_info_parsed r_reply_info;
01deead0	287	struct page *r_locked_page;
2f2dc053	288	int r_err;
c36d6414 JL	289	int r_num_caps;
c36d6414 JL	290	u32 r_readdir_offset;
2f2dc053	291
a319bf56	292	unsigned long r_timeout; /* optional. jiffies, 0 is "wait forever" */
2f2dc053 SW	293	unsigned long r_started; /* start time to measure timeout against */
	294	unsigned long r_request_started; /* start time for mds request only,
	295	used to measure lease durations */
	296
	297	/* link unsafe requests to parent directory, for fsync */
	298	struct inode *r_unsafe_dir;
	299	struct list_head r_unsafe_dir_item;
	300
68cd5b4b YZ	301	/* unsafe requests that modify the target inode */
	302	struct list_head r_unsafe_target_item;
	303
2f2dc053 SW	304	struct ceph_mds_session *r_session;
	305
	306	int r_attempts; /* resend attempts */
	307	int r_num_fwd; /* number of forward attempts */
2f2dc053	308	int r_resend_mds; /* mds to resend to next, if any*/
e55b71f8	309	u32 r_sent_on_mseq; /* cap mseq request was sent at*/
2f2dc053	310
2f2dc053 SW	311	struct list_head r_wait;
	312	struct completion r_completion;
	313	struct completion r_safe_completion;
	314	ceph_mds_request_callback_t r_callback;
9280be24	315	ceph_mds_request_wait_callback_t r_wait_for_completion;
2f2dc053	316	struct list_head r_unsafe_item; /* per-session unsafe list item */
2f2dc053	317
fdd4e158 YZ	318	long long r_dir_release_cnt;
	319	long long r_dir_ordered_cnt;
	320	int r_readdir_cache_idx;
2f2dc053 SW	321
2f2dc053 SW	322	struct ceph_cap_reservation r_caps_reservation;
2f2dc053 SW	323	};
2f2dc053 SW	324
10183a69 YZ	325	struct ceph_pool_perm {
10183a69 YZ	326	struct rb_node node;
10183a69	327	int perm;
7627151e	328	s64 pool;
779fe0fb YZ	329	size_t pool_ns_len;
779fe0fb YZ	330	char pool_ns[];
10183a69 YZ	331	};
10183a69 YZ	332
75c9627e YZ	333	struct ceph_snapid_map {
	334	struct rb_node node;
	335	struct list_head lru;
	336	atomic_t ref;
	337	u64 snap;
	338	dev_t dev;
	339	unsigned long last_used;
	340	};
	341
0c44a8e0 LH	342	/*
	343	* node for list of quotarealm inodes that are not visible from the filesystem
	344	* mountpoint, but required to handle, e.g. quotas.
	345	*/
	346	struct ceph_quotarealm_inode {
	347	struct rb_node node;
	348	u64 ino;
	349	unsigned long timeout; /* last time a lookup failed for this inode */
	350	struct mutex mutex;
	351	struct inode *inode;
	352	};
	353
3a3430af JL	354	struct cap_wait {
	355	struct list_head list;
	356	unsigned long ino;
	357	pid_t tgid;
	358	int need;
	359	int want;
	360	};
	361
2f2dc053 SW	362	/*
	363	* mds client state
	364	*/
	365	struct ceph_mds_client {
3d14c5d2	366	struct ceph_fs_client *fsc;
2f2dc053 SW	367	struct mutex mutex; /* all nested structures */
	368
	369	struct ceph_mdsmap *mdsmap;
f3c60c59 SW	370	struct completion safe_umount_waiters;
f3c60c59 SW	371	wait_queue_head_t session_close_wq;
2f2dc053	372	struct list_head waiting_for_map;
430afbad	373	int mdsmap_err;
2f2dc053 SW	374
2f2dc053 SW	375	struct ceph_mds_session *sessions; / NULL for mds if no session */
86d8f67b	376	atomic_t num_sessions;
2f2dc053 SW	377	int max_sessions; /* len of s_mds_sessions */
	378	int stopping; /* true if shutting down */
	379
d557c48d	380	atomic64_t quotarealms_count; /* # realms with quota */
0c44a8e0 LH	381	/*
	382	* We keep a list of inodes we don't see in the mountpoint but that we
	383	* need to track quota realms.
	384	*/
	385	struct rb_root quotarealms_inodes;
	386	struct mutex quotarealms_inodes_mutex;
d557c48d	387
2f2dc053 SW	388	/*
	389	* snap_rwsem will cover cap linkage into snaprealms, and
	390	* realm snap contexts. (later, we can do per-realm snap
	391	* contexts locks..) the empty list contains realms with no
	392	* references (implying they contain no inodes with caps) that
	393	* should be destroyed.
	394	*/
affbc19a	395	u64 last_snap_seq;
2f2dc053	396	struct rw_semaphore snap_rwsem;
a105f00c	397	struct rb_root snap_realms;
2f2dc053	398	struct list_head snap_empty;
81c5a148	399	int num_snap_realms;
2f2dc053 SW	400	spinlock_t snap_empty_lock; /* protect snap_empty */
	401
	402	u64 last_tid; /* most recent mds request */
e8a7b8b1 YZ	403	u64 oldest_tid; /* oldest incomplete mds request,
e8a7b8b1 YZ	404	excluding setfilelock requests */
44ca18f2	405	struct rb_root request_tree; /* pending mds requests */
2f2dc053 SW	406	struct delayed_work delayed_work; /* delayed work */
	407	unsigned long last_renew_caps; /* last time we renewed our caps */
	408	struct list_head cap_delay_list; /* caps with delayed release */
	409	spinlock_t cap_delay_lock; /* protects cap_delay_list */
	410	struct list_head snap_flush_list; /* cap_snaps ready to flush */
	411	spinlock_t snap_flush_lock;
	412
553adfd9	413	u64 last_cap_flush_tid;
e4500b5e	414	struct list_head cap_flush_list;
2f2dc053	415	struct list_head cap_dirty; /* inodes with dirty caps */
db354052	416	struct list_head cap_dirty_migrating; /* ...that are migration... */
2f2dc053 SW	417	int num_cap_flushing; /* # caps we are flushing */
	418	spinlock_t cap_dirty_lock; /* protects above items */
	419	wait_queue_head_t cap_flushing_wq;
	420
37c4efc1	421	struct work_struct cap_reclaim_work;
fe33032d	422	atomic_t cap_reclaim_pending;
37c4efc1	423
37151668 YS	424	/*
	425	* Cap reservations
	426	*
	427	* Maintain a global pool of preallocated struct ceph_caps, referenced
	428	* by struct ceph_caps_reservations. This ensures that we preallocate
	429	* memory needed to successfully process an MDS response. (If an MDS
	430	* sends us cap information and we fail to process it, we will have
	431	* problems due to the client and MDS being out of sync.)
	432	*
	433	* Reservations are 'owned' by a ceph_cap_reservation context.
	434	*/
	435	spinlock_t caps_list_lock;
	436	struct list_head caps_list; /* unused (reserved or
	437	unreserved) */
3a3430af	438	struct list_head cap_wait_list;
37151668 YS	439	int caps_total_count; /* total caps allocated */
37151668 YS	440	int caps_use_count; /* in use */
fe33032d	441	int caps_use_max; /* max used caps */
37151668 YS	442	int caps_reserve_count; /* unused, reserved */
	443	int caps_avail_count; /* unused, unreserved */
	444	int caps_min_count; /* keep at least this many
	445	(unreserved) */
37c4efc1 YZ	446	spinlock_t dentry_list_lock;
	447	struct list_head dentry_leases; /* fifo list */
	448	struct list_head dentry_dir_leases; /* lru list */
10183a69	449
75c9627e YZ	450	spinlock_t snapid_map_lock;
	451	struct rb_root snapid_map_tree;
	452	struct list_head snapid_map_lru;
	453
10183a69 YZ	454	struct rw_semaphore pool_perm_rwsem;
10183a69 YZ	455	struct rb_root pool_perm_tree;
717e6f28 YZ	456
717e6f28 YZ	457	char nodename[__NEW_UTS_LEN + 1];
2f2dc053 SW	458	};
	459
	460	extern const char *ceph_mds_op_name(int op);
	461
	462	extern struct ceph_mds_session *
	463	__ceph_lookup_mds_session(struct ceph_mds_client *, int mds);
	464
a687ecaf JS	465	extern const char *ceph_session_state_name(int s);
a687ecaf JS	466
5b3248c6 XL	467	extern struct ceph_mds_session *
5b3248c6 XL	468	ceph_get_mds_session(struct ceph_mds_session *s);
2f2dc053 SW	469	extern void ceph_put_mds_session(struct ceph_mds_session *s);
	470
	471	extern int ceph_send_msg_mds(struct ceph_mds_client *mdsc,
	472	struct ceph_msg *msg, int mds);
	473
3d14c5d2	474	extern int ceph_mdsc_init(struct ceph_fs_client *fsc);
2f2dc053	475	extern void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc);
48fec5d0	476	extern void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc);
3d14c5d2	477	extern void ceph_mdsc_destroy(struct ceph_fs_client *fsc);
2f2dc053 SW	478
	479	extern void ceph_mdsc_sync(struct ceph_mds_client *mdsc);
	480
167c9e35	481	extern void ceph_invalidate_dir_request(struct ceph_mds_request *req);
54008399 YZ	482	extern int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
54008399 YZ	483	struct inode *dir);
2f2dc053 SW	484	extern struct ceph_mds_request *
2f2dc053 SW	485	ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode);
86bda539 JL	486	extern int ceph_mdsc_submit_request(struct ceph_mds_client *mdsc,
	487	struct inode *dir,
	488	struct ceph_mds_request *req);
2f2dc053 SW	489	extern int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
	490	struct inode *dir,
	491	struct ceph_mds_request *req);
	492	static inline void ceph_mdsc_get_request(struct ceph_mds_request *req)
	493	{
153c8e6b SW	494	kref_get(&req->r_kref);
	495	}
	496	extern void ceph_mdsc_release_request(struct kref *kref);
	497	static inline void ceph_mdsc_put_request(struct ceph_mds_request *req)
	498	{
	499	kref_put(&req->r_kref, ceph_mdsc_release_request);
2f2dc053	500	}
2f2dc053	501
e3ec8d68 YZ	502	extern void __ceph_queue_cap_release(struct ceph_mds_session *session,
	503	struct ceph_cap *cap);
	504	extern void ceph_flush_cap_releases(struct ceph_mds_client *mdsc,
	505	struct ceph_mds_session *session);
37c4efc1	506	extern void ceph_queue_cap_reclaim_work(struct ceph_mds_client *mdsc);
fe33032d	507	extern void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr);
f5d77269 JL	508	extern int ceph_iterate_session_caps(struct ceph_mds_session *session,
	509	int (cb)(struct inode ,
	510	struct ceph_cap , void ),
	511	void *arg);
2f2dc053 SW	512	extern void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc);
2f2dc053 SW	513
f77f21bb JL	514	static inline void ceph_mdsc_free_path(char *path, int len)
	515	{
	516	if (path)
	517	__putname(path - (PATH_MAX - 1 - len));
	518	}
	519
2f2dc053 SW	520	extern char ceph_mdsc_build_path(struct dentry dentry, int plen, u64 base,
	521	int stop_on_nosnap);
	522
	523	extern void __ceph_mdsc_drop_dentry_lease(struct dentry *dentry);
	524	extern void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
2f2dc053 SW	525	struct dentry *dentry, char action,
	526	u32 seq);
	527
430afbad YZ	528	extern void ceph_mdsc_handle_mdsmap(struct ceph_mds_client *mdsc,
	529	struct ceph_msg *msg);
	530	extern void ceph_mdsc_handle_fsmap(struct ceph_mds_client *mdsc,
	531	struct ceph_msg *msg);
2f2dc053	532
5d72d13c YZ	533	extern struct ceph_mds_session *
5d72d13c YZ	534	ceph_mdsc_open_export_target_session(struct ceph_mds_client *mdsc, int target);
154f42c2 SW	535	extern void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
	536	struct ceph_mds_session *session);
	537
e30ee581 ZZ	538	extern int ceph_trim_caps(struct ceph_mds_client *mdsc,
	539	struct ceph_mds_session *session,
	540	int max_caps);
2f2dc053	541	#endif