smc_core.h 12.9 KB
Newer Older
1
/* SPDX-License-Identifier: GPL-2.0 */
2 3 4 5 6 7 8 9 10 11 12 13 14
/*
 * Shared Memory Communications over RDMA (SMC-R) and RoCE
 *
 *  Definitions for SMC Connections, Link Groups and Links
 *
 *  Copyright IBM Corp. 2016
 *
 *  Author(s):  Ursula Braun <ubraun@linux.vnet.ibm.com>
 */

#ifndef _SMC_CORE_H
#define _SMC_CORE_H

15
#include <linux/atomic.h>
16 17 18 19 20
#include <rdma/ib_verbs.h>

#include "smc.h"
#include "smc_ib.h"

U
Ursula Braun 已提交
21 22
#define SMC_RMBS_PER_LGR_MAX	255	/* max. # of RMBs per link group */

23 24 25
struct smc_lgr_list {			/* list of link group definition */
	struct list_head	list;
	spinlock_t		lock;	/* protects list of link groups */
26
	u32			num;	/* unique link group number */
27 28 29 30 31 32 33
};

enum smc_lgr_role {		/* possible roles of a link group */
	SMC_CLNT,	/* client */
	SMC_SERV	/* server */
};

34
enum smc_link_state {			/* possible states of a link */
35
	SMC_LNK_UNUSED,		/* link is unused */
36 37
	SMC_LNK_INACTIVE,	/* link is inactive */
	SMC_LNK_ACTIVATING,	/* link is being activated */
38
	SMC_LNK_ACTIVE,		/* link is active */
39 40
};

41 42 43 44 45 46
#define SMC_WR_BUF_SIZE		48	/* size of work request buffer */

struct smc_wr_buf {
	u8	raw[SMC_WR_BUF_SIZE];
};

47 48 49 50 51 52 53 54
#define SMC_WR_REG_MR_WAIT_TIME	(5 * HZ)/* wait time for ib_wr_reg_mr result */

enum smc_wr_reg_state {
	POSTED,		/* ib_wr_reg_mr request posted */
	CONFIRMED,	/* ib_wr_reg_mr response: successful */
	FAILED		/* ib_wr_reg_mr response: failure */
};

55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72
struct smc_rdma_sge {				/* sges for RDMA writes */
	struct ib_sge		wr_tx_rdma_sge[SMC_IB_MAX_SEND_SGE];
};

#define SMC_MAX_RDMA_WRITES	2		/* max. # of RDMA writes per
						 * message send
						 */

struct smc_rdma_sges {				/* sges per message send */
	struct smc_rdma_sge	tx_rdma_sge[SMC_MAX_RDMA_WRITES];
};

struct smc_rdma_wr {				/* work requests per message
						 * send
						 */
	struct ib_rdma_wr	wr_tx_rdma[SMC_MAX_RDMA_WRITES];
};

73 74 75
struct smc_link {
	struct smc_ib_device	*smcibdev;	/* ib-device */
	u8			ibport;		/* port - values 1 | 2 */
76 77 78
	struct ib_pd		*roce_pd;	/* IB protection domain,
						 * unique for every RoCE QP
						 */
79 80
	struct ib_qp		*roce_qp;	/* IB queue pair */
	struct ib_qp_attr	qp_attr;	/* IB queue pair attributes */
81 82 83 84

	struct smc_wr_buf	*wr_tx_bufs;	/* WR send payload buffers */
	struct ib_send_wr	*wr_tx_ibs;	/* WR send meta data */
	struct ib_sge		*wr_tx_sges;	/* WR send gather meta data */
85 86
	struct smc_rdma_sges	*wr_tx_rdma_sges;/*RDMA WRITE gather meta data*/
	struct smc_rdma_wr	*wr_tx_rdmas;	/* WR RDMA WRITE */
87
	struct smc_wr_tx_pend	*wr_tx_pends;	/* WR send waiting for CQE */
88
	struct completion	*wr_tx_compl;	/* WR send CQE completion */
89 90 91 92 93 94 95 96 97 98 99 100 101 102
	/* above four vectors have wr_tx_cnt elements and use the same index */
	dma_addr_t		wr_tx_dma_addr;	/* DMA address of wr_tx_bufs */
	atomic_long_t		wr_tx_id;	/* seq # of last sent WR */
	unsigned long		*wr_tx_mask;	/* bit mask of used indexes */
	u32			wr_tx_cnt;	/* number of WR send buffers */
	wait_queue_head_t	wr_tx_wait;	/* wait for free WR send buf */

	struct smc_wr_buf	*wr_rx_bufs;	/* WR recv payload buffers */
	struct ib_recv_wr	*wr_rx_ibs;	/* WR recv meta data */
	struct ib_sge		*wr_rx_sges;	/* WR recv scatter meta data */
	/* above three vectors have wr_rx_cnt elements and use the same index */
	dma_addr_t		wr_rx_dma_addr;	/* DMA address of wr_rx_bufs */
	u64			wr_rx_id;	/* seq # of last recv WR */
	u32			wr_rx_cnt;	/* number of WR recv buffers */
103
	unsigned long		wr_rx_tstamp;	/* jiffies when last buf rx */
104

105 106 107 108
	struct ib_reg_wr	wr_reg;		/* WR register memory region */
	wait_queue_head_t	wr_reg_wait;	/* wait for wr_reg result */
	enum smc_wr_reg_state	wr_reg_state;	/* state of wr_reg request */

109 110
	u8			gid[SMC_GID_SIZE];/* gid matching used vlan id*/
	u8			sgid_index;	/* gid index for vlan id      */
111 112 113 114 115 116
	u32			peer_qpn;	/* QP number of peer */
	enum ib_mtu		path_mtu;	/* used mtu */
	enum ib_mtu		peer_mtu;	/* mtu size of peer */
	u32			psn_initial;	/* QP tx initial packet seqno */
	u32			peer_psn;	/* QP rx initial packet seqno */
	u8			peer_mac[ETH_ALEN];	/* = gid[8:10||13:15] */
117
	u8			peer_gid[SMC_GID_SIZE];	/* gid of peer*/
U
Ursula Braun 已提交
118
	u8			link_id;	/* unique # within link group */
119
	u8			link_idx;	/* index in lgr link array */
K
Karsten Graul 已提交
120
	u8			link_is_asym;	/* is link asymmetric? */
121
	struct smc_link_group	*lgr;		/* parent link group */
122
	struct work_struct	link_down_wrk;	/* wrk to bring link down */
123 124

	enum smc_link_state	state;		/* state of link */
125 126 127
	struct delayed_work	llc_testlink_wrk; /* testlink worker */
	struct completion	llc_testlink_resp; /* wait for rx of testlink */
	int			llc_testlink_time; /* testlink interval */
128 129 130 131 132
};

/* For now we just allow one parallel link per link group. The SMC protocol
 * allows more (up to 8).
 */
133
#define SMC_LINKS_PER_LGR_MAX	3
134 135 136 137 138
#define SMC_SINGLE_LINK		0

#define SMC_FIRST_CONTACT	1		/* first contact to a peer */
#define SMC_REUSE_CONTACT	0		/* follow-on contact to a peer*/

U
Ursula Braun 已提交
139 140 141 142
/* tx/rx buffer list element for sndbufs list and rmbs list of a lgr */
struct smc_buf_desc {
	struct list_head	list;
	void			*cpu_addr;	/* virtual address of buffer */
143
	struct page		*pages;
144
	int			len;		/* length of buffer */
U
Ursula Braun 已提交
145
	u32			used;		/* currently used / unused */
146 147
	union {
		struct { /* SMC-R */
148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
			struct sg_table	sgt[SMC_LINKS_PER_LGR_MAX];
					/* virtual buffer */
			struct ib_mr	*mr_rx[SMC_LINKS_PER_LGR_MAX];
					/* for rmb only: memory region
					 * incl. rkey provided to peer
					 */
			u32		order;	/* allocation order */

			u8		is_conf_rkey;
					/* confirm_rkey done */
			u8		is_reg_mr[SMC_LINKS_PER_LGR_MAX];
					/* mem region registered */
			u8		is_map_ib[SMC_LINKS_PER_LGR_MAX];
					/* mem region mapped to lnk */
			u8		is_reg_err;
					/* buffer registration err */
164 165
		};
		struct { /* SMC-D */
166 167 168 169 170 171
			unsigned short	sba_idx;
					/* SBA index number */
			u64		token;
					/* DMB token number */
			dma_addr_t	dma_addr;
					/* DMA address */
172 173
		};
	};
U
Ursula Braun 已提交
174 175
};

176 177 178 179 180
struct smc_rtoken {				/* address/key of remote RMB */
	u64			dma_addr;
	u32			rkey;
};

U
Ursula Braun 已提交
181
#define SMC_LGR_ID_SIZE		4
182 183 184 185 186 187
#define SMC_BUF_MIN_SIZE	16384	/* minimum size of an RMB */
#define SMC_RMBE_SIZES		16	/* number of distinct RMBE sizes */
/* 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)
 */
U
Ursula Braun 已提交
188

189 190
struct smcd_dev;

191 192 193 194 195 196 197 198
enum smc_lgr_type {				/* redundancy state of lgr */
	SMC_LGR_NONE,			/* no active links, lgr to be deleted */
	SMC_LGR_SINGLE,			/* 1 active RNIC on each peer */
	SMC_LGR_SYMMETRIC,		/* 2 active RNICs on each peer */
	SMC_LGR_ASYMMETRIC_PEER,	/* local has 2, peer 1 active RNICs */
	SMC_LGR_ASYMMETRIC_LOCAL,	/* local has 1, peer 2 active RNICs */
};

199 200 201 202 203 204 205 206 207 208 209 210 211 212
enum smc_llc_flowtype {
	SMC_LLC_FLOW_NONE	= 0,
	SMC_LLC_FLOW_ADD_LINK	= 2,
	SMC_LLC_FLOW_DEL_LINK	= 4,
	SMC_LLC_FLOW_RKEY	= 6,
};

struct smc_llc_qentry;

struct smc_llc_flow {
	enum smc_llc_flowtype type;
	struct smc_llc_qentry *qentry;
};

213 214 215 216 217 218
struct smc_link_group {
	struct list_head	list;
	struct rb_root		conns_all;	/* connection tree */
	rwlock_t		conns_lock;	/* protects conns_all */
	unsigned int		conns_num;	/* current # of connections */
	unsigned short		vlan_id;	/* vlan id of link group */
U
Ursula Braun 已提交
219 220

	struct list_head	sndbufs[SMC_RMBE_SIZES];/* tx buffers */
221
	struct mutex		sndbufs_lock;	/* protects tx buffers */
U
Ursula Braun 已提交
222
	struct list_head	rmbs[SMC_RMBE_SIZES];	/* rx buffers */
223
	struct mutex		rmbs_lock;	/* protects rx buffers */
224

U
Ursula Braun 已提交
225
	u8			id[SMC_LGR_ID_SIZE];	/* unique lgr id */
226
	struct delayed_work	free_work;	/* delayed freeing of an lgr */
227
	struct work_struct	terminate_work;	/* abnormal lgr termination */
228 229
	u8			sync_err : 1;	/* lgr no longer fits to peer */
	u8			terminating : 1;/* lgr is terminating */
U
Ursula Braun 已提交
230 231
	u8			freefast : 1;	/* free worker scheduled fast */
	u8			freeing : 1;	/* lgr is being freed */
232 233 234 235 236 237 238 239 240 241 242 243 244

	bool			is_smcd;	/* SMC-R or SMC-D */
	union {
		struct { /* SMC-R */
			enum smc_lgr_role	role;
						/* client or server */
			struct smc_link		lnk[SMC_LINKS_PER_LGR_MAX];
						/* smc link */
			char			peer_systemid[SMC_SYSTEMID_LEN];
						/* unique system_id of peer */
			struct smc_rtoken	rtokens[SMC_RMBS_PER_LGR_MAX]
						[SMC_LINKS_PER_LGR_MAX];
						/* remote addr/key pairs */
245
			DECLARE_BITMAP(rtokens_used_mask, SMC_RMBS_PER_LGR_MAX);
246
						/* used rtoken elements */
247
			u8			next_link_id;
248 249
			enum smc_lgr_type	type;
						/* redundancy state */
250 251
			u8			pnet_id[SMC_MAX_PNETID_LEN + 1];
						/* pnet id of this lgr */
252 253 254 255
			struct list_head	llc_event_q;
						/* queue for llc events */
			spinlock_t		llc_event_q_lock;
						/* protects llc_event_q */
256 257
			struct mutex		llc_conf_mutex;
						/* protects lgr reconfig. */
258
			struct work_struct	llc_add_link_work;
259
			struct work_struct	llc_del_link_work;
260 261
			struct work_struct	llc_event_work;
						/* llc event worker */
262 263 264 265 266 267 268 269 270 271
			wait_queue_head_t	llc_waiter;
						/* w4 next llc event */
			struct smc_llc_flow	llc_flow_lcl;
						/* llc local control field */
			struct smc_llc_flow	llc_flow_rmt;
						/* llc remote control field */
			struct smc_llc_qentry	*delayed_event;
						/* arrived when flow active */
			spinlock_t		llc_flow_lock;
						/* protects llc flow */
272 273
			int			llc_testlink_time;
						/* link keep alive time */
274 275 276 277 278 279
		};
		struct { /* SMC-D */
			u64			peer_gid;
						/* Peer GID (remote) */
			struct smcd_dev		*smcd;
						/* ISM device for VLAN reg. */
280 281
			u8			peer_shutdown : 1;
						/* peer triggered shutdownn */
282 283
		};
	};
284 285
};

286 287 288 289 290 291
struct smc_clc_msg_local;

struct smc_init_info {
	u8			is_smcd;
	unsigned short		vlan_id;
	int			srv_first_contact;
292
	int			cln_first_contact;
293 294 295 296 297 298 299 300 301 302 303
	/* SMC-R */
	struct smc_clc_msg_local *ib_lcl;
	struct smc_ib_device	*ib_dev;
	u8			ib_gid[SMC_GID_SIZE];
	u8			ib_port;
	u32			ib_clcqpn;
	/* SMC-D */
	u64			ism_gid;
	struct smcd_dev		*ism_dev;
};

304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
/* Find the connection associated with the given alert token in the link group.
 * To use rbtrees we have to implement our own search core.
 * Requires @conns_lock
 * @token	alert token to search for
 * @lgr		 link group to search in
 * Returns connection associated with token if found, NULL otherwise.
 */
static inline struct smc_connection *smc_lgr_find_conn(
	u32 token, struct smc_link_group *lgr)
{
	struct smc_connection *res = NULL;
	struct rb_node *node;

	node = lgr->conns_all.rb_node;
	while (node) {
		struct smc_connection *cur = rb_entry(node,
					struct smc_connection, alert_node);

		if (cur->alert_token_local > token) {
			node = node->rb_left;
		} else {
			if (cur->alert_token_local < token) {
				node = node->rb_right;
			} else {
				res = cur;
				break;
			}
		}
	}

	return res;
}

337 338 339 340 341 342 343 344
/* returns true if the specified link is usable */
static inline bool smc_link_usable(struct smc_link *lnk)
{
	if (lnk->state == SMC_LNK_UNUSED || lnk->state == SMC_LNK_INACTIVE)
		return false;
	return true;
}

U
Ursula Braun 已提交
345 346
struct smc_sock;
struct smc_clc_msg_accept_confirm;
347
struct smc_clc_msg_local;
U
Ursula Braun 已提交
348

349
void smc_lgr_cleanup_early(struct smc_connection *conn);
350
void smc_lgr_terminate_sched(struct smc_link_group *lgr);
351
void smcr_port_add(struct smc_ib_device *smcibdev, u8 ibport);
352
void smcr_port_err(struct smc_ib_device *smcibdev, u8 ibport);
H
Hans Wippel 已提交
353 354
void smc_smcd_terminate(struct smcd_dev *dev, u64 peer_gid,
			unsigned short vlan);
355
void smc_smcd_terminate_all(struct smcd_dev *dev);
356
void smc_smcr_terminate_all(struct smc_ib_device *smcibdev);
357
int smc_buf_create(struct smc_sock *smc, bool is_smcd);
358
int smc_uncompress_bufsize(u8 compressed);
359
int smc_rmb_rtoken_handling(struct smc_connection *conn, struct smc_link *link,
360
			    struct smc_clc_msg_accept_confirm *clc);
361 362
int smc_rtoken_add(struct smc_link *lnk, __be64 nw_vaddr, __be32 nw_rkey);
int smc_rtoken_delete(struct smc_link *lnk, __be32 nw_rkey);
363 364 365 366
void smc_rtoken_set(struct smc_link_group *lgr, int link_idx, int link_idx_new,
		    __be32 nw_rkey_known, __be64 nw_vaddr, __be32 nw_rkey);
void smc_rtoken_set2(struct smc_link_group *lgr, int rtok_idx, int link_id,
		     __be64 nw_vaddr, __be32 nw_rkey);
367 368 369 370
void smc_sndbuf_sync_sg_for_cpu(struct smc_connection *conn);
void smc_sndbuf_sync_sg_for_device(struct smc_connection *conn);
void smc_rmb_sync_sg_for_cpu(struct smc_connection *conn);
void smc_rmb_sync_sg_for_device(struct smc_connection *conn);
371
int smc_vlan_by_tcpsk(struct socket *clcsock, struct smc_init_info *ini);
372

373
void smc_conn_free(struct smc_connection *conn);
374
int smc_conn_create(struct smc_sock *smc, struct smc_init_info *ini);
375
void smc_lgr_schedule_free_work_fast(struct smc_link_group *lgr);
376
int smc_core_init(void);
377
void smc_core_exit(void);
378

379 380
int smcr_link_init(struct smc_link_group *lgr, struct smc_link *lnk,
		   u8 link_idx, struct smc_init_info *ini);
381
void smcr_link_clear(struct smc_link *lnk);
382 383
int smcr_buf_map_lgr(struct smc_link *lnk);
int smcr_buf_reg_lgr(struct smc_link *lnk);
K
Karsten Graul 已提交
384 385 386
void smcr_lgr_set_type(struct smc_link_group *lgr, enum smc_lgr_type new_type);
void smcr_lgr_set_type_asym(struct smc_link_group *lgr,
			    enum smc_lgr_type new_type, int asym_lnk_idx);
387
int smcr_link_reg_rmb(struct smc_link *link, struct smc_buf_desc *rmb_desc);
388 389
struct smc_link *smc_switch_conns(struct smc_link_group *lgr,
				  struct smc_link *from_lnk, bool is_dev_err);
390 391 392
void smcr_link_down_cond(struct smc_link *lnk);
void smcr_link_down_cond_sched(struct smc_link *lnk);

393 394
static inline struct smc_link_group *smc_get_lgr(struct smc_link *link)
{
395
	return link->lgr;
396
}
397
#endif