[mirror_ubuntu-bionic-kernel.git] / net / smc / smc.h

/*
 *  Shared Memory Communications over RDMA (SMC-R) and RoCE
 *
 *  Definitions for the SMC module (socket related)
 *
 *  Copyright IBM Corp. 2016
 *
 *  Author(s):  Ursula Braun <ubraun@linux.vnet.ibm.com>
 */
#ifndef __SMC_H
#define __SMC_H

#include <linux/socket.h>
#include <linux/types.h>
#include <linux/compiler.h> /* __aligned */
#include <net/sock.h>

#include "smc_ib.h"

#define SMCPROTO_SMC		0	/* SMC protocol */

#define SMC_MAX_PORTS		2	/* Max # of ports */

extern struct proto smc_proto;

#ifdef ATOMIC64_INIT
#define KERNEL_HAS_ATOMIC64
#endif

enum smc_state {		/* possible states of an SMC socket */
	SMC_ACTIVE	= 1,
	SMC_INIT	= 2,
	SMC_CLOSED	= 7,
	SMC_LISTEN	= 10,
	/* normal close */
	SMC_PEERCLOSEWAIT1	= 20,
	SMC_PEERCLOSEWAIT2	= 21,
	SMC_APPFINCLOSEWAIT	= 24,
	SMC_APPCLOSEWAIT1	= 22,
	SMC_APPCLOSEWAIT2	= 23,
	SMC_PEERFINCLOSEWAIT	= 25,
	/* abnormal close */
	SMC_PEERABORTWAIT	= 26,
	SMC_PROCESSABORT	= 27,
};

struct smc_link_group;

struct smc_wr_rx_hdr {	/* common prefix part of LLC and CDC to demultiplex */
	u8			type;
} __aligned(1);

struct smc_cdc_conn_state_flags {
#if defined(__BIG_ENDIAN_BITFIELD)
	u8	peer_done_writing : 1;	/* Sending done indicator */
	u8	peer_conn_closed : 1;	/* Peer connection closed indicator */
	u8	peer_conn_abort : 1;	/* Abnormal close indicator */
	u8	reserved : 5;
#elif defined(__LITTLE_ENDIAN_BITFIELD)
	u8	reserved : 5;
	u8	peer_conn_abort : 1;
	u8	peer_conn_closed : 1;
	u8	peer_done_writing : 1;
#endif
};

struct smc_cdc_producer_flags {
#if defined(__BIG_ENDIAN_BITFIELD)
	u8	write_blocked : 1;	/* Writing Blocked, no rx buf space */
	u8	urg_data_pending : 1;	/* Urgent Data Pending */
	u8	urg_data_present : 1;	/* Urgent Data Present */
	u8	cons_curs_upd_req : 1;	/* cursor update requested */
	u8	failover_validation : 1;/* message replay due to failover */
	u8	reserved : 3;
#elif defined(__LITTLE_ENDIAN_BITFIELD)
	u8	reserved : 3;
	u8	failover_validation : 1;
	u8	cons_curs_upd_req : 1;
	u8	urg_data_present : 1;
	u8	urg_data_pending : 1;
	u8	write_blocked : 1;
#endif
};

/* in host byte order */
union smc_host_cursor {	/* SMC cursor - an offset in an RMBE */
	struct {
		u16	reserved;
		u16	wrap;		/* window wrap sequence number */
		u32	count;		/* cursor (= offset) part */
	};
#ifdef KERNEL_HAS_ATOMIC64
	atomic64_t		acurs;	/* for atomic processing */
#else
	u64			acurs;	/* for atomic processing */
#endif
} __aligned(8);

/* in host byte order, except for flag bitfields in network byte order */
struct smc_host_cdc_msg {		/* Connection Data Control message */
	struct smc_wr_rx_hdr		common; /* .type = 0xFE */
	u8				len;	/* length = 44 */
	u16				seqno;	/* connection seq # */
	u32				token;	/* alert_token */
	union smc_host_cursor		prod;		/* producer cursor */
	union smc_host_cursor		cons;		/* consumer cursor,
							 * piggy backed "ack"
							 */
	struct smc_cdc_producer_flags	prod_flags;	/* conn. tx/rx status */
	struct smc_cdc_conn_state_flags	conn_state_flags; /* peer conn. status*/
	u8				reserved[18];
} __aligned(8);

struct smc_connection {
	struct rb_node		alert_node;
	struct smc_link_group	*lgr;		/* link group of connection */
	u32			alert_token_local; /* unique conn. id */
	u8			peer_conn_idx;	/* from tcp handshake */
	int			peer_rmbe_size;	/* size of peer rx buffer */
	atomic_t		peer_rmbe_space;/* remaining free bytes in peer
						 * rmbe
						 */
	int			rtoken_idx;	/* idx to peer RMB rkey/addr */

	struct smc_buf_desc	*sndbuf_desc;	/* send buffer descriptor */
	int			sndbuf_size;	/* sndbuf size <== sock wmem */
	struct smc_buf_desc	*rmb_desc;	/* RMBE descriptor */
	int			rmbe_size;	/* RMBE size <== sock rmem */
	int			rmbe_size_short;/* compressed notation */
	int			rmbe_update_limit;
						/* lower limit for consumer
						 * cursor update
						 */

	struct smc_host_cdc_msg	local_tx_ctrl;	/* host byte order staging
						 * buffer for CDC msg send
						 * .prod cf. TCP snd_nxt
						 * .cons cf. TCP sends ack
						 */
	union smc_host_cursor	tx_curs_prep;	/* tx - prepared data
						 * snd_max..wmem_alloc
						 */
	union smc_host_cursor	tx_curs_sent;	/* tx - sent data
						 * snd_nxt ?
						 */
	union smc_host_cursor	tx_curs_fin;	/* tx - confirmed by peer
						 * snd-wnd-begin ?
						 */
	atomic_t		sndbuf_space;	/* remaining space in sndbuf */
	u16			tx_cdc_seq;	/* sequence # for CDC send */
	spinlock_t		send_lock;	/* protect wr_sends */
	struct work_struct	tx_work;	/* retry of smc_cdc_msg_send */

	struct smc_host_cdc_msg	local_rx_ctrl;	/* filled during event_handl.
						 * .prod cf. TCP rcv_nxt
						 * .cons cf. TCP snd_una
						 */
	union smc_host_cursor	rx_curs_confirmed; /* confirmed to peer
						    * source of snd_una ?
						    */
	atomic_t		bytes_to_rcv;	/* arrived data,
						 * not yet received
						 */
#ifndef KERNEL_HAS_ATOMIC64
	spinlock_t		acurs_lock;	/* protect cursors */
#endif
	struct work_struct	close_work;	/* peer sent some closing */
};

struct smc_sock {				/* smc sock container */
	struct sock		sk;
	struct socket		*clcsock;	/* internal tcp socket */
	struct smc_connection	conn;		/* smc connection */
	struct sockaddr		*addr;		/* inet connect address */
	struct smc_sock		*listen_smc;	/* listen parent */
	struct work_struct	tcp_listen_work;/* handle tcp socket accepts */
	struct work_struct	smc_listen_work;/* prepare new accept socket */
	struct list_head	accept_q;	/* sockets to be accepted */
	spinlock_t		accept_q_lock;	/* protects accept_q */
	struct delayed_work	sock_put_work;	/* final socket freeing */
	bool			use_fallback;	/* fallback to tcp */
	u8			wait_close_tx_prepared : 1;
						/* shutdown wr or close
						 * started, waiting for unsent
						 * data to be sent
						 */
};

static inline struct smc_sock *smc_sk(const struct sock *sk)
{
	return (struct smc_sock *)sk;
}

#define SMC_SYSTEMID_LEN		8

extern u8	local_systemid[SMC_SYSTEMID_LEN]; /* unique system identifier */

/* convert an u32 value into network byte order, store it into a 3 byte field */
static inline void hton24(u8 *net, u32 host)
{
	__be32 t;

	t = cpu_to_be32(host);
	memcpy(net, ((u8 *)&t) + 1, 3);
}

/* convert a received 3 byte field into host byte order*/
static inline u32 ntoh24(u8 *net)
{
	__be32 t = 0;

	memcpy(((u8 *)&t) + 1, net, 3);
	return be32_to_cpu(t);
}

#define SMC_BUF_MIN_SIZE 16384		/* minimum size of an RMB */

#define SMC_RMBE_SIZES	16	/* number of distinct sizes for an RMBE */
/* theoretically, the RFC states that largest size would be 512K,
 * i.e. compressed 5 and thus 6 sizes (0..5), despite
 * struct smc_clc_msg_accept_confirm.rmbe_size being a 4 bit value (0..15)
 */

/* convert the RMB size into the compressed notation - minimum 16K.
 * In contrast to plain ilog2, this rounds towards the next power of 2,
 * so the socket application gets at least its desired sndbuf / rcvbuf size.
 */
static inline u8 smc_compress_bufsize(int size)
{
	u8 compressed;

	if (size <= SMC_BUF_MIN_SIZE)
		return 0;

	size = (size - 1) >> 14;
	compressed = ilog2(size) + 1;
	if (compressed >= SMC_RMBE_SIZES)
		compressed = SMC_RMBE_SIZES - 1;
	return compressed;
}

/* convert the RMB size from compressed notation into integer */
static inline int smc_uncompress_bufsize(u8 compressed)
{
	u32 size;

	size = 0x00000001 << (((int)compressed) + 14);
	return (int)size;
}

#ifdef CONFIG_XFRM
static inline bool using_ipsec(struct smc_sock *smc)
{
	return (smc->clcsock->sk->sk_policy[0] ||
		smc->clcsock->sk->sk_policy[1]) ? 1 : 0;
}
#else
static inline bool using_ipsec(struct smc_sock *smc)
{
	return 0;
}
#endif

struct smc_clc_msg_local;

int smc_netinfo_by_tcpsk(struct socket *clcsock, __be32 *subnet,
			 u8 *prefix_len);
void smc_conn_free(struct smc_connection *conn);
int smc_conn_create(struct smc_sock *smc, __be32 peer_in_addr,
		    struct smc_ib_device *smcibdev, u8 ibport,
		    struct smc_clc_msg_local *lcl, int srv_first_contact);
struct sock *smc_accept_dequeue(struct sock *parent, struct socket *new_sock);
void smc_close_non_accepted(struct sock *sk);

#endif	/* __SMC_H */
Commit	Line	Data
ac713874 UB	1	/*
	2	* Shared Memory Communications over RDMA (SMC-R) and RoCE
	3	*
	4	* Definitions for the SMC module (socket related)
	5	*
	6	* Copyright IBM Corp. 2016
	7	*
	8	* Author(s): Ursula Braun <ubraun@linux.vnet.ibm.com>
	9	*/
	10	#ifndef __SMC_H
	11	#define __SMC_H
	12
	13	#include <linux/socket.h>
	14	#include <linux/types.h>
f38ba179	15	#include <linux/compiler.h> /* __aligned */
ac713874 UB	16	#include <net/sock.h>
ac713874 UB	17
0cfdd8f9 UB	18	#include "smc_ib.h"
0cfdd8f9 UB	19
ac713874 UB	20	#define SMCPROTO_SMC 0 /* SMC protocol */
ac713874 UB	21
a4cf0443 UB	22	#define SMC_MAX_PORTS 2 /* Max # of ports */
a4cf0443 UB	23
f16a7dd5 UB	24	extern struct proto smc_proto;
f16a7dd5 UB	25
5f08318f UB	26	#ifdef ATOMIC64_INIT
	27	#define KERNEL_HAS_ATOMIC64
	28	#endif
	29
ac713874 UB	30	enum smc_state { /* possible states of an SMC socket */
	31	SMC_ACTIVE = 1,
	32	SMC_INIT = 2,
	33	SMC_CLOSED = 7,
	34	SMC_LISTEN = 10,
b38d7324 UB	35	/* normal close */
	36	SMC_PEERCLOSEWAIT1 = 20,
	37	SMC_PEERCLOSEWAIT2 = 21,
	38	SMC_APPFINCLOSEWAIT = 24,
	39	SMC_APPCLOSEWAIT1 = 22,
	40	SMC_APPCLOSEWAIT2 = 23,
	41	SMC_PEERFINCLOSEWAIT = 25,
	42	/* abnormal close */
	43	SMC_PEERABORTWAIT = 26,
	44	SMC_PROCESSABORT = 27,
ac713874 UB	45	};
ac713874 UB	46
0cfdd8f9 UB	47	struct smc_link_group;
0cfdd8f9 UB	48
f38ba179 UB	49	struct smc_wr_rx_hdr { /* common prefix part of LLC and CDC to demultiplex */
	50	u8 type;
	51	} __aligned(1);
	52
5f08318f UB	53	struct smc_cdc_conn_state_flags {
	54	#if defined(__BIG_ENDIAN_BITFIELD)
	55	u8 peer_done_writing : 1; /* Sending done indicator */
	56	u8 peer_conn_closed : 1; /* Peer connection closed indicator */
	57	u8 peer_conn_abort : 1; /* Abnormal close indicator */
	58	u8 reserved : 5;
	59	#elif defined(__LITTLE_ENDIAN_BITFIELD)
	60	u8 reserved : 5;
	61	u8 peer_conn_abort : 1;
	62	u8 peer_conn_closed : 1;
	63	u8 peer_done_writing : 1;
	64	#endif
	65	};
	66
	67	struct smc_cdc_producer_flags {
	68	#if defined(__BIG_ENDIAN_BITFIELD)
	69	u8 write_blocked : 1; /* Writing Blocked, no rx buf space */
	70	u8 urg_data_pending : 1; /* Urgent Data Pending */
	71	u8 urg_data_present : 1; /* Urgent Data Present */
	72	u8 cons_curs_upd_req : 1; /* cursor update requested */
	73	u8 failover_validation : 1;/* message replay due to failover */
	74	u8 reserved : 3;
	75	#elif defined(__LITTLE_ENDIAN_BITFIELD)
	76	u8 reserved : 3;
	77	u8 failover_validation : 1;
	78	u8 cons_curs_upd_req : 1;
	79	u8 urg_data_present : 1;
	80	u8 urg_data_pending : 1;
	81	u8 write_blocked : 1;
	82	#endif
	83	};
	84
	85	/* in host byte order */
	86	union smc_host_cursor { /* SMC cursor - an offset in an RMBE */
	87	struct {
	88	u16 reserved;
	89	u16 wrap; /* window wrap sequence number */
	90	u32 count; /* cursor (= offset) part */
	91	};
	92	#ifdef KERNEL_HAS_ATOMIC64
	93	atomic64_t acurs; /* for atomic processing */
	94	#else
	95	u64 acurs; /* for atomic processing */
	96	#endif
	97	} __aligned(8);
	98
	99	/* in host byte order, except for flag bitfields in network byte order */
	100	struct smc_host_cdc_msg { /* Connection Data Control message */
	101	struct smc_wr_rx_hdr common; /* .type = 0xFE */
	102	u8 len; /* length = 44 */
	103	u16 seqno; /* connection seq # */
	104	u32 token; /* alert_token */
	105	union smc_host_cursor prod; /* producer cursor */
	106	union smc_host_cursor cons; /* consumer cursor,
	107	* piggy backed "ack"
	108	*/
	109	struct smc_cdc_producer_flags prod_flags; /* conn. tx/rx status */
	110	struct smc_cdc_conn_state_flags conn_state_flags; /* peer conn. status*/
	111	u8 reserved[18];
	112	} __aligned(8);
	113
0cfdd8f9 UB	114	struct smc_connection {
	115	struct rb_node alert_node;
	116	struct smc_link_group lgr; / link group of connection */
	117	u32 alert_token_local; /* unique conn. id */
	118	u8 peer_conn_idx; /* from tcp handshake */
cd6851f3 UB	119	int peer_rmbe_size; /* size of peer rx buffer */
	120	atomic_t peer_rmbe_space;/* remaining free bytes in peer
	121	* rmbe
	122	*/
bd4ad577	123	int rtoken_idx; /* idx to peer RMB rkey/addr */
cd6851f3 UB	124
	125	struct smc_buf_desc sndbuf_desc; / send buffer descriptor */
	126	int sndbuf_size; /* sndbuf size <== sock wmem */
	127	struct smc_buf_desc rmb_desc; / RMBE descriptor */
	128	int rmbe_size; /* RMBE size <== sock rmem */
	129	int rmbe_size_short;/* compressed notation */
952310cc UB	130	int rmbe_update_limit;
	131	/* lower limit for consumer
	132	* cursor update
	133	*/
5f08318f UB	134
	135	struct smc_host_cdc_msg local_tx_ctrl; /* host byte order staging
	136	* buffer for CDC msg send
	137	* .prod cf. TCP snd_nxt
	138	* .cons cf. TCP sends ack
	139	*/
	140	union smc_host_cursor tx_curs_prep; /* tx - prepared data
	141	* snd_max..wmem_alloc
	142	*/
	143	union smc_host_cursor tx_curs_sent; /* tx - sent data
	144	* snd_nxt ?
	145	*/
	146	union smc_host_cursor tx_curs_fin; /* tx - confirmed by peer
	147	* snd-wnd-begin ?
	148	*/
	149	atomic_t sndbuf_space; /* remaining space in sndbuf */
	150	u16 tx_cdc_seq; /* sequence # for CDC send */
	151	spinlock_t send_lock; /* protect wr_sends */
e6727f39	152	struct work_struct tx_work; /* retry of smc_cdc_msg_send */
5f08318f UB	153
	154	struct smc_host_cdc_msg local_rx_ctrl; /* filled during event_handl.
	155	* .prod cf. TCP rcv_nxt
	156	* .cons cf. TCP snd_una
	157	*/
	158	union smc_host_cursor rx_curs_confirmed; /* confirmed to peer
	159	* source of snd_una ?
	160	*/
	161	atomic_t bytes_to_rcv; /* arrived data,
	162	* not yet received
	163	*/
	164	#ifndef KERNEL_HAS_ATOMIC64
	165	spinlock_t acurs_lock; /* protect cursors */
	166	#endif
46c28dbd	167	struct work_struct close_work; /* peer sent some closing */
0cfdd8f9 UB	168	};
0cfdd8f9 UB	169
ac713874 UB	170	struct smc_sock { /* smc sock container */
	171	struct sock sk;
	172	struct socket clcsock; / internal tcp socket */
0cfdd8f9	173	struct smc_connection conn; /* smc connection */
a046d57d UB	174	struct sockaddr addr; / inet connect address */
	175	struct smc_sock listen_smc; / listen parent */
	176	struct work_struct tcp_listen_work;/* handle tcp socket accepts */
	177	struct work_struct smc_listen_work;/* prepare new accept socket */
	178	struct list_head accept_q; /* sockets to be accepted */
	179	spinlock_t accept_q_lock; /* protects accept_q */
b38d7324	180	struct delayed_work sock_put_work; /* final socket freeing */
ac713874	181	bool use_fallback; /* fallback to tcp */
b38d7324 UB	182	u8 wait_close_tx_prepared : 1;
	183	/* shutdown wr or close
	184	* started, waiting for unsent
	185	* data to be sent
	186	*/
ac713874 UB	187	};
	188
	189	static inline struct smc_sock smc_sk(const struct sock sk)
	190	{
	191	return (struct smc_sock *)sk;
	192	}
	193
a4cf0443 UB	194	#define SMC_SYSTEMID_LEN 8
	195
	196	extern u8 local_systemid[SMC_SYSTEMID_LEN]; /* unique system identifier */
	197
0cfdd8f9 UB	198	/* convert an u32 value into network byte order, store it into a 3 byte field */
	199	static inline void hton24(u8 *net, u32 host)
	200	{
	201	__be32 t;
	202
	203	t = cpu_to_be32(host);
	204	memcpy(net, ((u8 *)&t) + 1, 3);
	205	}
	206
	207	/* convert a received 3 byte field into host byte order*/
	208	static inline u32 ntoh24(u8 *net)
	209	{
	210	__be32 t = 0;
	211
	212	memcpy(((u8 *)&t) + 1, net, 3);
	213	return be32_to_cpu(t);
	214	}
	215
cd6851f3 UB	216	#define SMC_BUF_MIN_SIZE 16384 /* minimum size of an RMB */
	217
	218	#define SMC_RMBE_SIZES 16 /* number of distinct sizes for an RMBE */
	219	/* theoretically, the RFC states that largest size would be 512K,
	220	* i.e. compressed 5 and thus 6 sizes (0..5), despite
	221	* struct smc_clc_msg_accept_confirm.rmbe_size being a 4 bit value (0..15)
	222	*/
	223
	224	/* convert the RMB size into the compressed notation - minimum 16K.
	225	* In contrast to plain ilog2, this rounds towards the next power of 2,
	226	* so the socket application gets at least its desired sndbuf / rcvbuf size.
	227	*/
	228	static inline u8 smc_compress_bufsize(int size)
	229	{
	230	u8 compressed;
	231
	232	if (size <= SMC_BUF_MIN_SIZE)
	233	return 0;
	234
	235	size = (size - 1) >> 14;
	236	compressed = ilog2(size) + 1;
	237	if (compressed >= SMC_RMBE_SIZES)
	238	compressed = SMC_RMBE_SIZES - 1;
	239	return compressed;
	240	}
	241
	242	/* convert the RMB size from compressed notation into integer */
	243	static inline int smc_uncompress_bufsize(u8 compressed)
	244	{
	245	u32 size;
	246
	247	size = 0x00000001 << (((int)compressed) + 14);
	248	return (int)size;
	249	}
	250
a046d57d UB	251	#ifdef CONFIG_XFRM
	252	static inline bool using_ipsec(struct smc_sock *smc)
	253	{
	254	return (smc->clcsock->sk->sk_policy[0] \|\|
	255	smc->clcsock->sk->sk_policy[1]) ? 1 : 0;
	256	}
	257	#else
	258	static inline bool using_ipsec(struct smc_sock *smc)
	259	{
	260	return 0;
	261	}
	262	#endif
	263
0cfdd8f9 UB	264	struct smc_clc_msg_local;
0cfdd8f9 UB	265
a046d57d UB	266	int smc_netinfo_by_tcpsk(struct socket clcsock, __be32 subnet,
a046d57d UB	267	u8 *prefix_len);
0cfdd8f9 UB	268	void smc_conn_free(struct smc_connection *conn);
	269	int smc_conn_create(struct smc_sock *smc, __be32 peer_in_addr,
	270	struct smc_ib_device *smcibdev, u8 ibport,
	271	struct smc_clc_msg_local *lcl, int srv_first_contact);
b38d7324 UB	272	struct sock smc_accept_dequeue(struct sock parent, struct socket *new_sock);
b38d7324 UB	273	void smc_close_non_accepted(struct sock *sk);
a046d57d	274
ac713874	275	#endif /* __SMC_H */