[linux.git] / include / linux / ceph / messenger.h

#ifndef __FS_CEPH_MESSENGER_H
#define __FS_CEPH_MESSENGER_H

#include <linux/kref.h>
#include <linux/mutex.h>
#include <linux/net.h>
#include <linux/radix-tree.h>
#include <linux/uio.h>
#include <linux/workqueue.h>

#include "types.h"
#include "buffer.h"

struct ceph_msg;
struct ceph_connection;

/*
 * Ceph defines these callbacks for handling connection events.
 */
struct ceph_connection_operations {
	struct ceph_connection *(*get)(struct ceph_connection *);
	void (*put)(struct ceph_connection *);

	/* handle an incoming message. */
	void (*dispatch) (struct ceph_connection *con, struct ceph_msg *m);

	/* authorize an outgoing connection */
	struct ceph_auth_handshake *(*get_authorizer) (
				struct ceph_connection *con,
			       int *proto, int force_new);
	int (*verify_authorizer_reply) (struct ceph_connection *con, int len);
	int (*invalidate_authorizer)(struct ceph_connection *con);

	/* protocol version mismatch */
	void (*bad_proto) (struct ceph_connection *con);

	/* there was some error on the socket (disconnect, whatever) */
	void (*fault) (struct ceph_connection *con);

	/* a remote host as terminated a message exchange session, and messages
	 * we sent (or they tried to send us) may be lost. */
	void (*peer_reset) (struct ceph_connection *con);

	struct ceph_msg * (*alloc_msg) (struct ceph_connection *con,
					struct ceph_msg_header *hdr,
					int *skip);
};

/* use format string %s%d */
#define ENTITY_NAME(n) ceph_entity_type_name((n).type), le64_to_cpu((n).num)

struct ceph_messenger {
	struct ceph_entity_inst inst;    /* my name+address */
	struct ceph_entity_addr my_enc_addr;

	bool nocrc;

	/*
	 * the global_seq counts connections i (attempt to) initiate
	 * in order to disambiguate certain connect race conditions.
	 */
	u32 global_seq;
	spinlock_t global_seq_lock;

	u32 supported_features;
	u32 required_features;
};

/*
 * a single message.  it contains a header (src, dest, message type, etc.),
 * footer (crc values, mainly), a "front" message body, and possibly a
 * data payload (stored in some number of pages).
 */
struct ceph_msg {
	struct ceph_msg_header hdr;	/* header */
	struct ceph_msg_footer footer;	/* footer */
	struct kvec front;              /* unaligned blobs of message */
	struct ceph_buffer *middle;
	struct page **pages;            /* data payload.  NOT OWNER. */
	unsigned nr_pages;              /* size of page array */
	unsigned page_alignment;        /* io offset in first page */
	struct ceph_pagelist *pagelist; /* instead of pages */
	struct list_head list_head;
	struct kref kref;
	struct bio  *bio;		/* instead of pages/pagelist */
	struct bio  *bio_iter;		/* bio iterator */
	int bio_seg;			/* current bio segment */
	struct ceph_pagelist *trail;	/* the trailing part of the data */
	bool front_is_vmalloc;
	bool more_to_follow;
	bool needs_out_seq;
	int front_max;
	unsigned long ack_stamp;        /* tx: when we were acked */

	struct ceph_msgpool *pool;
};

struct ceph_msg_pos {
	int page, page_pos;  /* which page; offset in page */
	int data_pos;        /* offset in data payload */
	bool did_page_crc;   /* true if we've calculated crc for current page */
};

/* ceph connection fault delay defaults, for exponential backoff */
#define BASE_DELAY_INTERVAL	(HZ/2)
#define MAX_DELAY_INTERVAL	(5 * 60 * HZ)

/*
 * ceph_connection state bit flags
 */
#define LOSSYTX         0  /* we can close channel or drop messages on errors */
#define CONNECTING	1
#define NEGOTIATING	2
#define KEEPALIVE_PENDING      3
#define WRITE_PENDING	4  /* we have data ready to send */
#define STANDBY		8  /* no outgoing messages, socket closed.  we keep
			    * the ceph_connection around to maintain shared
			    * state with the peer. */
#define CLOSED		10 /* we've closed the connection */
#define SOCK_CLOSED	11 /* socket state changed to closed */
#define OPENING         13 /* open connection w/ (possibly new) peer */
#define DEAD            14 /* dead, about to kfree */
#define BACKOFF         15

/*
 * A single connection with another host.
 *
 * We maintain a queue of outgoing messages, and some session state to
 * ensure that we can preserve the lossless, ordered delivery of
 * messages in the case of a TCP disconnect.
 */
struct ceph_connection {
	void *private;
	atomic_t nref;

	const struct ceph_connection_operations *ops;

	struct ceph_messenger *msgr;
	struct socket *sock;
	unsigned long state;	/* connection state (see flags above) */
	const char *error_msg;  /* error message, if any */

	struct ceph_entity_addr peer_addr; /* peer address */
	struct ceph_entity_name peer_name; /* peer name */
	struct ceph_entity_addr peer_addr_for_me;
	unsigned peer_features;
	u32 connect_seq;      /* identify the most recent connection
				 attempt for this connection, client */
	u32 peer_global_seq;  /* peer's global seq for this connection */

	int auth_retry;       /* true if we need a newer authorizer */
	void *auth_reply_buf;   /* where to put the authorizer reply */
	int auth_reply_buf_len;

	struct mutex mutex;

	/* out queue */
	struct list_head out_queue;
	struct list_head out_sent;   /* sending or sent but unacked */
	u64 out_seq;		     /* last message queued for send */

	u64 in_seq, in_seq_acked;  /* last message received, acked */

	/* connection negotiation temps */
	char in_banner[CEPH_BANNER_MAX_LEN];
	union {
		struct {  /* outgoing connection */
			struct ceph_msg_connect out_connect;
			struct ceph_msg_connect_reply in_reply;
		};
		struct {  /* incoming */
			struct ceph_msg_connect in_connect;
			struct ceph_msg_connect_reply out_reply;
		};
	};
	struct ceph_entity_addr actual_peer_addr;

	/* message out temps */
	struct ceph_msg *out_msg;        /* sending message (== tail of
					    out_sent) */
	bool out_msg_done;
	struct ceph_msg_pos out_msg_pos;

	struct kvec out_kvec[8],         /* sending header/footer data */
		*out_kvec_cur;
	int out_kvec_left;   /* kvec's left in out_kvec */
	int out_skip;        /* skip this many bytes */
	int out_kvec_bytes;  /* total bytes left */
	bool out_kvec_is_msg; /* kvec refers to out_msg */
	int out_more;        /* there is more data after the kvecs */
	__le64 out_temp_ack; /* for writing an ack */

	/* message in temps */
	struct ceph_msg_header in_hdr;
	struct ceph_msg *in_msg;
	struct ceph_msg_pos in_msg_pos;
	u32 in_front_crc, in_middle_crc, in_data_crc;  /* calculated crc */

	char in_tag;         /* protocol control byte */
	int in_base_pos;     /* bytes read */
	__le64 in_temp_ack;  /* for reading an ack */

	struct delayed_work work;	    /* send|recv work */
	unsigned long       delay;          /* current delay interval */
};


extern const char *ceph_pr_addr(const struct sockaddr_storage *ss);
extern int ceph_parse_ips(const char *c, const char *end,
			  struct ceph_entity_addr *addr,
			  int max_count, int *count);


extern int ceph_msgr_init(void);
extern void ceph_msgr_exit(void);
extern void ceph_msgr_flush(void);

extern struct ceph_messenger *ceph_messenger_create(
	struct ceph_entity_addr *myaddr,
	u32 features, u32 required);
extern void ceph_messenger_destroy(struct ceph_messenger *);

extern void ceph_con_init(struct ceph_messenger *msgr,
			  struct ceph_connection *con);
extern void ceph_con_open(struct ceph_connection *con,
			  struct ceph_entity_addr *addr);
extern bool ceph_con_opened(struct ceph_connection *con);
extern void ceph_con_close(struct ceph_connection *con);
extern void ceph_con_send(struct ceph_connection *con, struct ceph_msg *msg);
extern void ceph_con_revoke(struct ceph_connection *con, struct ceph_msg *msg);
extern void ceph_con_revoke_message(struct ceph_connection *con,
				  struct ceph_msg *msg);
extern void ceph_con_keepalive(struct ceph_connection *con);
extern struct ceph_connection *ceph_con_get(struct ceph_connection *con);
extern void ceph_con_put(struct ceph_connection *con);

extern struct ceph_msg *ceph_msg_new(int type, int front_len, gfp_t flags,
				     bool can_fail);
extern void ceph_msg_kfree(struct ceph_msg *m);


static inline struct ceph_msg *ceph_msg_get(struct ceph_msg *msg)
{
	kref_get(&msg->kref);
	return msg;
}
extern void ceph_msg_last_put(struct kref *kref);
static inline void ceph_msg_put(struct ceph_msg *msg)
{
	kref_put(&msg->kref, ceph_msg_last_put);
}

extern void ceph_msg_dump(struct ceph_msg *msg);

#endif
Commit	Line	Data
31b8006e SW	1	#ifndef __FS_CEPH_MESSENGER_H
	2	#define __FS_CEPH_MESSENGER_H
	3
c2e552e7	4	#include <linux/kref.h>
31b8006e SW	5	#include <linux/mutex.h>
	6	#include <linux/net.h>
	7	#include <linux/radix-tree.h>
	8	#include <linux/uio.h>
31b8006e SW	9	#include <linux/workqueue.h>
	10
	11	#include "types.h"
	12	#include "buffer.h"
	13
	14	struct ceph_msg;
	15	struct ceph_connection;
	16
31b8006e SW	17	/*
	18	* Ceph defines these callbacks for handling connection events.
	19	*/
	20	struct ceph_connection_operations {
	21	struct ceph_connection (get)(struct ceph_connection *);
	22	void (put)(struct ceph_connection );
	23
	24	/* handle an incoming message. */
	25	void (dispatch) (struct ceph_connection con, struct ceph_msg *m);
	26
4e7a5dcd	27	/* authorize an outgoing connection */
a3530df3 AE	28	struct ceph_auth_handshake (get_authorizer) (
a3530df3 AE	29	struct ceph_connection *con,
8f43fb53	30	int *proto, int force_new);
4e7a5dcd	31	int (verify_authorizer_reply) (struct ceph_connection con, int len);
9bd2e6f8	32	int (invalidate_authorizer)(struct ceph_connection con);
4e7a5dcd	33
31b8006e SW	34	/* protocol version mismatch */
	35	void (bad_proto) (struct ceph_connection con);
	36
	37	/* there was some error on the socket (disconnect, whatever) */
	38	void (fault) (struct ceph_connection con);
	39
	40	/* a remote host as terminated a message exchange session, and messages
	41	* we sent (or they tried to send us) may be lost. */
	42	void (peer_reset) (struct ceph_connection con);
	43
	44	struct ceph_msg * (alloc_msg) (struct ceph_connection con,
2450418c YS	45	struct ceph_msg_header *hdr,
2450418c YS	46	int *skip);
31b8006e SW	47	};
31b8006e SW	48
31b8006e	49	/* use format string %s%d */
ca9d93a2	50	#define ENTITY_NAME(n) ceph_entity_type_name((n).type), le64_to_cpu((n).num)
31b8006e SW	51
	52	struct ceph_messenger {
	53	struct ceph_entity_inst inst; /* my name+address */
63f2d211	54	struct ceph_entity_addr my_enc_addr;
31b8006e SW	55
	56	bool nocrc;
	57
	58	/*
	59	* the global_seq counts connections i (attempt to) initiate
	60	* in order to disambiguate certain connect race conditions.
	61	*/
	62	u32 global_seq;
	63	spinlock_t global_seq_lock;
3d14c5d2 YS	64
	65	u32 supported_features;
	66	u32 required_features;
31b8006e SW	67	};
	68
	69	/*
	70	* a single message. it contains a header (src, dest, message type, etc.),
	71	* footer (crc values, mainly), a "front" message body, and possibly a
	72	* data payload (stored in some number of pages).
	73	*/
	74	struct ceph_msg {
	75	struct ceph_msg_header hdr; /* header */
	76	struct ceph_msg_footer footer; /* footer */
	77	struct kvec front; /* unaligned blobs of message */
	78	struct ceph_buffer *middle;
	79	struct page *pages; / data payload. NOT OWNER. */
	80	unsigned nr_pages; /* size of page array */
c5c6b19d	81	unsigned page_alignment; /* io offset in first page */
58bb3b37	82	struct ceph_pagelist pagelist; / instead of pages */
31b8006e	83	struct list_head list_head;
c2e552e7	84	struct kref kref;
68b4476b YS	85	struct bio bio; / instead of pages/pagelist */
	86	struct bio bio_iter; / bio iterator */
	87	int bio_seg; /* current bio segment */
	88	struct ceph_pagelist trail; / the trailing part of the data */
31b8006e SW	89	bool front_is_vmalloc;
31b8006e SW	90	bool more_to_follow;
e84346b7	91	bool needs_out_seq;
31b8006e	92	int front_max;
4cf9d544	93	unsigned long ack_stamp; /* tx: when we were acked */
31b8006e SW	94
	95	struct ceph_msgpool *pool;
	96	};
	97
	98	struct ceph_msg_pos {
	99	int page, page_pos; /* which page; offset in page */
	100	int data_pos; /* offset in data payload */
bca064d2	101	bool did_page_crc; /* true if we've calculated crc for current page */
31b8006e SW	102	};
	103
	104	/* ceph connection fault delay defaults, for exponential backoff */
	105	#define BASE_DELAY_INTERVAL (HZ/2)
	106	#define MAX_DELAY_INTERVAL (5 * 60 * HZ)
	107
	108	/*
	109	* ceph_connection state bit flags
31b8006e SW	110	*/
31b8006e SW	111	#define LOSSYTX 0 /* we can close channel or drop messages on errors */
eed0ef2c SW	112	#define CONNECTING 1
eed0ef2c SW	113	#define NEGOTIATING 2
31b8006e SW	114	#define KEEPALIVE_PENDING 3
31b8006e SW	115	#define WRITE_PENDING 4 /* we have data ready to send */
31b8006e SW	116	#define STANDBY 8 /* no outgoing messages, socket closed. we keep
	117	* the ceph_connection around to maintain shared
	118	* state with the peer. */
	119	#define CLOSED 10 /* we've closed the connection */
	120	#define SOCK_CLOSED 11 /* socket state changed to closed */
31b8006e SW	121	#define OPENING 13 /* open connection w/ (possibly new) peer */
31b8006e SW	122	#define DEAD 14 /* dead, about to kfree */
60bf8bf8	123	#define BACKOFF 15
31b8006e SW	124
	125	/*
	126	* A single connection with another host.
	127	*
	128	* We maintain a queue of outgoing messages, and some session state to
	129	* ensure that we can preserve the lossless, ordered delivery of
	130	* messages in the case of a TCP disconnect.
	131	*/
	132	struct ceph_connection {
	133	void *private;
	134	atomic_t nref;
	135
	136	const struct ceph_connection_operations *ops;
	137
	138	struct ceph_messenger *msgr;
	139	struct socket *sock;
	140	unsigned long state; /* connection state (see flags above) */
	141	const char error_msg; / error message, if any */
	142
	143	struct ceph_entity_addr peer_addr; /* peer address */
	144	struct ceph_entity_name peer_name; /* peer name */
	145	struct ceph_entity_addr peer_addr_for_me;
aba558e2	146	unsigned peer_features;
31b8006e SW	147	u32 connect_seq; /* identify the most recent connection
	148	attempt for this connection, client */
	149	u32 peer_global_seq; /* peer's global seq for this connection */
	150
4e7a5dcd SW	151	int auth_retry; /* true if we need a newer authorizer */
	152	void auth_reply_buf; / where to put the authorizer reply */
	153	int auth_reply_buf_len;
	154
ec302645 SW	155	struct mutex mutex;
ec302645 SW	156
31b8006e	157	/* out queue */
31b8006e SW	158	struct list_head out_queue;
	159	struct list_head out_sent; /* sending or sent but unacked */
	160	u64 out_seq; /* last message queued for send */
31b8006e SW	161
	162	u64 in_seq, in_seq_acked; /* last message received, acked */
	163
	164	/* connection negotiation temps */
	165	char in_banner[CEPH_BANNER_MAX_LEN];
	166	union {
	167	struct { /* outgoing connection */
	168	struct ceph_msg_connect out_connect;
	169	struct ceph_msg_connect_reply in_reply;
	170	};
	171	struct { /* incoming */
	172	struct ceph_msg_connect in_connect;
	173	struct ceph_msg_connect_reply out_reply;
	174	};
	175	};
	176	struct ceph_entity_addr actual_peer_addr;
	177
	178	/* message out temps */
	179	struct ceph_msg out_msg; / sending message (== tail of
	180	out_sent) */
c86a2930	181	bool out_msg_done;
31b8006e SW	182	struct ceph_msg_pos out_msg_pos;
	183
	184	struct kvec out_kvec[8], /* sending header/footer data */
	185	*out_kvec_cur;
	186	int out_kvec_left; /* kvec's left in out_kvec */
	187	int out_skip; /* skip this many bytes */
	188	int out_kvec_bytes; /* total bytes left */
	189	bool out_kvec_is_msg; /* kvec refers to out_msg */
	190	int out_more; /* there is more data after the kvecs */
	191	__le64 out_temp_ack; /* for writing an ack */
	192
	193	/* message in temps */
	194	struct ceph_msg_header in_hdr;
	195	struct ceph_msg *in_msg;
	196	struct ceph_msg_pos in_msg_pos;
	197	u32 in_front_crc, in_middle_crc, in_data_crc; /* calculated crc */
	198
	199	char in_tag; /* protocol control byte */
	200	int in_base_pos; /* bytes read */
	201	__le64 in_temp_ack; /* for reading an ack */
	202
	203	struct delayed_work work; /* send\|recv work */
	204	unsigned long delay; /* current delay interval */
	205	};
	206
	207
3d14c5d2	208	extern const char ceph_pr_addr(const struct sockaddr_storage ss);
31b8006e SW	209	extern int ceph_parse_ips(const char c, const char end,
	210	struct ceph_entity_addr *addr,
	211	int max_count, int *count);
	212
	213
	214	extern int ceph_msgr_init(void);
	215	extern void ceph_msgr_exit(void);
a922d38f	216	extern void ceph_msgr_flush(void);
31b8006e SW	217
31b8006e SW	218	extern struct ceph_messenger *ceph_messenger_create(
3d14c5d2 YS	219	struct ceph_entity_addr *myaddr,
3d14c5d2 YS	220	u32 features, u32 required);
31b8006e SW	221	extern void ceph_messenger_destroy(struct ceph_messenger *);
	222
	223	extern void ceph_con_init(struct ceph_messenger *msgr,
	224	struct ceph_connection *con);
31b8006e SW	225	extern void ceph_con_open(struct ceph_connection *con,
31b8006e SW	226	struct ceph_entity_addr *addr);
87b315a5	227	extern bool ceph_con_opened(struct ceph_connection *con);
31b8006e SW	228	extern void ceph_con_close(struct ceph_connection *con);
	229	extern void ceph_con_send(struct ceph_connection con, struct ceph_msg msg);
	230	extern void ceph_con_revoke(struct ceph_connection con, struct ceph_msg msg);
0d59ab81 YS	231	extern void ceph_con_revoke_message(struct ceph_connection *con,
0d59ab81 YS	232	struct ceph_msg *msg);
31b8006e SW	233	extern void ceph_con_keepalive(struct ceph_connection *con);
	234	extern struct ceph_connection ceph_con_get(struct ceph_connection con);
	235	extern void ceph_con_put(struct ceph_connection *con);
	236
b61c2763 SW	237	extern struct ceph_msg *ceph_msg_new(int type, int front_len, gfp_t flags,
b61c2763 SW	238	bool can_fail);
31b8006e SW	239	extern void ceph_msg_kfree(struct ceph_msg *m);
31b8006e SW	240
31b8006e SW	241
	242	static inline struct ceph_msg ceph_msg_get(struct ceph_msg msg)
	243	{
c2e552e7	244	kref_get(&msg->kref);
31b8006e SW	245	return msg;
31b8006e SW	246	}
c2e552e7 SW	247	extern void ceph_msg_last_put(struct kref *kref);
	248	static inline void ceph_msg_put(struct ceph_msg *msg)
	249	{
	250	kref_put(&msg->kref, ceph_msg_last_put);
	251	}
31b8006e	252
9ec7cab1 SW	253	extern void ceph_msg_dump(struct ceph_msg *msg);
9ec7cab1 SW	254
31b8006e	255	#endif