Merge branch 'smc-link-layer-control-enhancements'

Ursula Braun says:

====================
net/smc: Link Layer Control enhancements

here is a series of smc patches enabling SMC communication with peers
supporting more than one link per link group.

The first three patches are preparing code cleanups.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

net/smc/af_smc.c

@@ -7,7 +7,6 @@
  *  applicable with RoCE-cards only
  *
  *  Initial restrictions:
- *    - non-blocking connect postponed
  *    - IPv6 support postponed
  *    - support for alternate links postponed
  *    - partial support for non-blocking sockets only
@@ -24,7 +23,6 @@
 
 #include <linux/module.h>
 #include <linux/socket.h>
-#include <linux/inetdevice.h>
 #include <linux/workqueue.h>
 #include <linux/in.h>
 #include <linux/sched/signal.h>
@@ -273,46 +271,7 @@ static void smc_copy_sock_settings_to_smc(struct smc_sock *smc)
 	smc_copy_sock_settings(&smc->sk, smc->clcsock->sk, SK_FLAGS_CLC_TO_SMC);
 }
 
-/* determine subnet and mask of internal TCP socket */
-int smc_netinfo_by_tcpsk(struct socket *clcsock,
-			 __be32 *subnet, u8 *prefix_len)
-{
-	struct dst_entry *dst = sk_dst_get(clcsock->sk);
-	struct in_device *in_dev;
-	struct sockaddr_in addr;
-	int rc = -ENOENT;
-
-	if (!dst) {
-		rc = -ENOTCONN;
-		goto out;
-	}
-	if (!dst->dev) {
-		rc = -ENODEV;
-		goto out_rel;
-	}
-
-	/* get address to which the internal TCP socket is bound */
-	kernel_getsockname(clcsock, (struct sockaddr *)&addr);
-	/* analyze IPv4 specific data of net_device belonging to TCP socket */
-	rcu_read_lock();
-	in_dev = __in_dev_get_rcu(dst->dev);
-	for_ifa(in_dev) {
-		if (!inet_ifa_match(addr.sin_addr.s_addr, ifa))
-			continue;
-		*prefix_len = inet_mask_len(ifa->ifa_mask);
-		*subnet = ifa->ifa_address & ifa->ifa_mask;
-		rc = 0;
-		break;
-	} endfor_ifa(in_dev);
-	rcu_read_unlock();
-
-out_rel:
-	dst_release(dst);
-out:
-	return rc;
-}
-
-static int smc_clnt_conf_first_link(struct smc_sock *smc, union ib_gid *gid)
+static int smc_clnt_conf_first_link(struct smc_sock *smc)
 {
 	struct smc_link_group *lgr = smc->conn.lgr;
 	struct smc_link *link;
@@ -332,6 +291,9 @@ static int smc_clnt_conf_first_link(struct smc_sock *smc, union ib_gid *gid)
 		return rc;
 	}
 
+	if (link->llc_confirm_rc)
+		return SMC_CLC_DECL_RMBE_EC;
+
 	rc = smc_ib_modify_qp_rts(link);
 	if (rc)
 		return SMC_CLC_DECL_INTERR;
@@ -346,11 +308,33 @@ static int smc_clnt_conf_first_link(struct smc_sock *smc, union ib_gid *gid)
 	/* send CONFIRM LINK response over RoCE fabric */
 	rc = smc_llc_send_confirm_link(link,
 				       link->smcibdev->mac[link->ibport - 1],
-				       gid, SMC_LLC_RESP);
+				       &link->smcibdev->gid[link->ibport - 1],
+				       SMC_LLC_RESP);
 	if (rc < 0)
 		return SMC_CLC_DECL_TCL;
 
+	/* receive ADD LINK request from server over RoCE fabric */
+	rest = wait_for_completion_interruptible_timeout(&link->llc_add,
+							 SMC_LLC_WAIT_TIME);
+	if (rest <= 0) {
+		struct smc_clc_msg_decline dclc;
+
+		rc = smc_clc_wait_msg(smc, &dclc, sizeof(dclc),
+				      SMC_CLC_DECLINE);
 		return rc;
+	}
+
+	/* send add link reject message, only one link supported for now */
+	rc = smc_llc_send_add_link(link,
+				   link->smcibdev->mac[link->ibport - 1],
+				   &link->smcibdev->gid[link->ibport - 1],
+				   SMC_LLC_RESP);
+	if (rc < 0)
+		return SMC_CLC_DECL_TCL;
+
+	link->state = SMC_LNK_ACTIVE;
+
+	return 0;
 }
 
 static void smc_conn_save_peer_info(struct smc_sock *smc,
@@ -372,19 +356,9 @@ static void smc_link_save_peer_info(struct smc_link *link,
 	link->peer_mtu = clc->qp_mtu;
 }
 
-static void smc_lgr_forget(struct smc_link_group *lgr)
-{
-	spin_lock_bh(&smc_lgr_list.lock);
-	/* do not use this link group for new connections */
-	if (!list_empty(&lgr->list))
-		list_del_init(&lgr->list);
-	spin_unlock_bh(&smc_lgr_list.lock);
-}
-
 /* setup for RDMA connection of client */
 static int smc_connect_rdma(struct smc_sock *smc)
 {
-	struct sockaddr_in *inaddr = (struct sockaddr_in *)smc->addr;
 	struct smc_clc_msg_accept_confirm aclc;
 	int local_contact = SMC_FIRST_CONTACT;
 	struct smc_ib_device *smcibdev;
@@ -438,8 +412,8 @@ static int smc_connect_rdma(struct smc_sock *smc)
 
 	srv_first_contact = aclc.hdr.flag;
 	mutex_lock(&smc_create_lgr_pending);
-	local_contact = smc_conn_create(smc, inaddr->sin_addr.s_addr, smcibdev,
-					ibport, &aclc.lcl, srv_first_contact);
+	local_contact = smc_conn_create(smc, smcibdev, ibport, &aclc.lcl,
+					srv_first_contact);
 	if (local_contact < 0) {
 		rc = local_contact;
 		if (rc == -ENOMEM)
@@ -498,8 +472,7 @@ static int smc_connect_rdma(struct smc_sock *smc)
 
 	if (local_contact == SMC_FIRST_CONTACT) {
 		/* QP confirmation over RoCE fabric */
-		reason_code = smc_clnt_conf_first_link(
-			smc, &smcibdev->gid[ibport - 1]);
+		reason_code = smc_clnt_conf_first_link(smc);
 		if (reason_code < 0) {
 			rc = reason_code;
 			goto out_err_unlock;
@@ -558,7 +531,6 @@ static int smc_connect(struct socket *sock, struct sockaddr *addr,
 		goto out_err;
 	if (addr->sa_family != AF_INET)
 		goto out_err;
-	smc->addr = addr;	/* needed for nonblocking connect */
 
 	lock_sock(sk);
 	switch (sk->sk_state) {
@@ -748,9 +720,34 @@ static int smc_serv_conf_first_link(struct smc_sock *smc)
 
 		rc = smc_clc_wait_msg(smc, &dclc, sizeof(dclc),
 				      SMC_CLC_DECLINE);
+		return rc;
 	}
 
+	if (link->llc_confirm_resp_rc)
+		return SMC_CLC_DECL_RMBE_EC;
+
+	/* send ADD LINK request to client over the RoCE fabric */
+	rc = smc_llc_send_add_link(link,
+				   link->smcibdev->mac[link->ibport - 1],
+				   &link->smcibdev->gid[link->ibport - 1],
+				   SMC_LLC_REQ);
+	if (rc < 0)
+		return SMC_CLC_DECL_TCL;
+
+	/* receive ADD LINK response from client over the RoCE fabric */
+	rest = wait_for_completion_interruptible_timeout(&link->llc_add_resp,
+							 SMC_LLC_WAIT_TIME);
+	if (rest <= 0) {
+		struct smc_clc_msg_decline dclc;
+
+		rc = smc_clc_wait_msg(smc, &dclc, sizeof(dclc),
+				      SMC_CLC_DECLINE);
 		return rc;
+	}
+
+	link->state = SMC_LNK_ACTIVE;
+
+	return 0;
 }
 
 /* setup for RDMA connection of server */
@@ -766,7 +763,6 @@ static void smc_listen_work(struct work_struct *work)
 	struct sock *newsmcsk = &new_smc->sk;
 	struct smc_clc_msg_proposal *pclc;
 	struct smc_ib_device *smcibdev;
-	struct sockaddr_in peeraddr;
 	u8 buf[SMC_CLC_MAX_LEN];
 	struct smc_link *link;
 	int reason_code = 0;
@@ -808,7 +804,7 @@ static void smc_listen_work(struct work_struct *work)
 	}
 
 	/* determine subnet and mask from internal TCP socket */
-	rc = smc_netinfo_by_tcpsk(newclcsock, &subnet, &prefix_len);
+	rc = smc_clc_netinfo_by_tcpsk(newclcsock, &subnet, &prefix_len);
 	if (rc) {
 		reason_code = SMC_CLC_DECL_CNFERR; /* configuration error */
 		goto decline_rdma;
@@ -822,13 +818,10 @@ static void smc_listen_work(struct work_struct *work)
 		goto decline_rdma;
 	}
 
-	/* get address of the peer connected to the internal TCP socket */
-	kernel_getpeername(newclcsock, (struct sockaddr *)&peeraddr);
-
 	/* allocate connection / link group */
 	mutex_lock(&smc_create_lgr_pending);
-	local_contact = smc_conn_create(new_smc, peeraddr.sin_addr.s_addr,
-					smcibdev, ibport, &pclc->lcl, 0);
+	local_contact = smc_conn_create(new_smc, smcibdev, ibport, &pclc->lcl,
+					0);
 	if (local_contact < 0) {
 		rc = local_contact;
 		if (rc == -ENOMEM)

net/smc/smc.h

@@ -172,7 +172,6 @@ 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 */
@@ -263,10 +262,8 @@ static inline bool using_ipsec(struct smc_sock *smc)
 
 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,
+int smc_conn_create(struct smc_sock *smc,
 		    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);

net/smc/smc_clc.c

@@ -11,6 +11,7 @@
  */
 
 #include <linux/in.h>
+#include <linux/inetdevice.h>
 #include <linux/if_ether.h>
 #include <linux/sched/signal.h>
 
@@ -22,6 +23,9 @@
 #include "smc_clc.h"
 #include "smc_ib.h"
 
+/* eye catcher "SMCR" EBCDIC for CLC messages */
+static const char SMC_EYECATCHER[4] = {'\xe2', '\xd4', '\xc3', '\xd9'};
+
 /* check if received message has a correct header length and contains valid
  * heading and trailing eyecatchers
  */
@@ -70,6 +74,45 @@ static bool smc_clc_msg_hdr_valid(struct smc_clc_msg_hdr *clcm)
 	return true;
 }
 
+/* determine subnet and mask of internal TCP socket */
+int smc_clc_netinfo_by_tcpsk(struct socket *clcsock,
+			     __be32 *subnet, u8 *prefix_len)
+{
+	struct dst_entry *dst = sk_dst_get(clcsock->sk);
+	struct in_device *in_dev;
+	struct sockaddr_in addr;
+	int rc = -ENOENT;
+
+	if (!dst) {
+		rc = -ENOTCONN;
+		goto out;
+	}
+	if (!dst->dev) {
+		rc = -ENODEV;
+		goto out_rel;
+	}
+
+	/* get address to which the internal TCP socket is bound */
+	kernel_getsockname(clcsock, (struct sockaddr *)&addr);
+	/* analyze IPv4 specific data of net_device belonging to TCP socket */
+	rcu_read_lock();
+	in_dev = __in_dev_get_rcu(dst->dev);
+	for_ifa(in_dev) {
+		if (!inet_ifa_match(addr.sin_addr.s_addr, ifa))
+			continue;
+		*prefix_len = inet_mask_len(ifa->ifa_mask);
+		*subnet = ifa->ifa_address & ifa->ifa_mask;
+		rc = 0;
+		break;
+	} endfor_ifa(in_dev);
+	rcu_read_unlock();
+
+out_rel:
+	dst_release(dst);
+out:
+	return rc;
+}
+
 /* Wait for data on the tcp-socket, analyze received data
  * Returns:
  * 0 if success and it was not a decline that we received.
@@ -211,7 +254,7 @@ int smc_clc_send_proposal(struct smc_sock *smc,
 
 	memset(&pclc_prfx, 0, sizeof(pclc_prfx));
 	/* determine subnet and mask from internal TCP socket */
-	rc = smc_netinfo_by_tcpsk(smc->clcsock, &pclc_prfx.outgoing_subnet,
+	rc = smc_clc_netinfo_by_tcpsk(smc->clcsock, &pclc_prfx.outgoing_subnet,
 				      &pclc_prfx.prefix_len);
 	if (rc)
 		return SMC_CLC_DECL_CNFERR; /* configuration error */

net/smc/smc_clc.h

@@ -22,9 +22,6 @@
 #define SMC_CLC_CONFIRM		0x03
 #define SMC_CLC_DECLINE		0x04
 
-/* eye catcher "SMCR" EBCDIC for CLC messages */
-static const char SMC_EYECATCHER[4] = {'\xe2', '\xd4', '\xc3', '\xd9'};
-
 #define SMC_CLC_V1		0x1		/* SMC version                */
 #define CLC_WAIT_TIME		(6 * HZ)	/* max. wait time on clcsock  */
 #define SMC_CLC_DECL_MEM	0x01010000  /* insufficient memory resources  */
@@ -36,6 +33,7 @@ static const char SMC_EYECATCHER[4] = {'\xe2', '\xd4', '\xc3', '\xd9'};
 #define SMC_CLC_DECL_INTERR	0x99990000  /* internal error                 */
 #define SMC_CLC_DECL_TCL	0x02040000  /* timeout w4 QP confirm          */
 #define SMC_CLC_DECL_SEND	0x07000000  /* sending problem                */
+#define SMC_CLC_DECL_RMBE_EC	0x08000000  /* peer has eyecatcher in RMBE    */
 
 struct smc_clc_msg_hdr {	/* header1 of clc messages */
 	u8 eyecatcher[4];	/* eye catcher */
@@ -124,9 +122,8 @@ smc_clc_proposal_get_prefix(struct smc_clc_msg_proposal *pclc)
 	       ((u8 *)pclc + sizeof(*pclc) + ntohs(pclc->iparea_offset));
 }
 
-struct smc_sock;
-struct smc_ib_device;
-
+int smc_clc_netinfo_by_tcpsk(struct socket *clcsock, __be32 *subnet,
+			     u8 *prefix_len);
 int smc_clc_wait_msg(struct smc_sock *smc, void *buf, int buflen,
 		     u8 expected_type);
 int smc_clc_send_decline(struct smc_sock *smc, u32 peer_diag_info);

net/smc/smc_core.c

@@ -144,7 +144,7 @@ static void smc_lgr_free_work(struct work_struct *work)
 }
 
 /* create a new SMC link group */
-static int smc_lgr_create(struct smc_sock *smc, __be32 peer_in_addr,
+static int smc_lgr_create(struct smc_sock *smc,
 			  struct smc_ib_device *smcibdev, u8 ibport,
 			  char *peer_systemid, unsigned short vlan_id)
 {
@@ -161,7 +161,6 @@ static int smc_lgr_create(struct smc_sock *smc, __be32 peer_in_addr,
 	}
 	lgr->role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
 	lgr->sync_err = false;
-	lgr->daddr = peer_in_addr;
 	memcpy(lgr->peer_systemid, peer_systemid, SMC_SYSTEMID_LEN);
 	lgr->vlan_id = vlan_id;
 	rwlock_init(&lgr->sndbufs_lock);
@@ -177,6 +176,7 @@ static int smc_lgr_create(struct smc_sock *smc, __be32 peer_in_addr,
 
 	lnk = &lgr->lnk[SMC_SINGLE_LINK];
 	/* initialize link */
+	lnk->state = SMC_LNK_ACTIVATING;
 	lnk->smcibdev = smcibdev;
 	lnk->ibport = ibport;
 	lnk->path_mtu = smcibdev->pattr[ibport - 1].active_mtu;
@@ -198,6 +198,8 @@ static int smc_lgr_create(struct smc_sock *smc, __be32 peer_in_addr,
 		goto destroy_qp;
 	init_completion(&lnk->llc_confirm);
 	init_completion(&lnk->llc_confirm_resp);
+	init_completion(&lnk->llc_add);
+	init_completion(&lnk->llc_add_resp);
 
 	smc->conn.lgr = lgr;
 	rwlock_init(&lgr->conns_lock);
@@ -306,6 +308,15 @@ void smc_lgr_free(struct smc_link_group *lgr)
 	kfree(lgr);
 }
 
+void smc_lgr_forget(struct smc_link_group *lgr)
+{
+	spin_lock_bh(&smc_lgr_list.lock);
+	/* do not use this link group for new connections */
+	if (!list_empty(&lgr->list))
+		list_del_init(&lgr->list);
+	spin_unlock_bh(&smc_lgr_list.lock);
+}
+
 /* terminate linkgroup abnormally */
 void smc_lgr_terminate(struct smc_link_group *lgr)
 {
@@ -313,15 +324,7 @@ void smc_lgr_terminate(struct smc_link_group *lgr)
 	struct smc_sock *smc;
 	struct rb_node *node;
 
-	spin_lock_bh(&smc_lgr_list.lock);
-	if (list_empty(&lgr->list)) {
-		/* termination already triggered */
-		spin_unlock_bh(&smc_lgr_list.lock);
-		return;
-	}
-	/* do not use this link group for new connections */
-	list_del_init(&lgr->list);
-	spin_unlock_bh(&smc_lgr_list.lock);
+	smc_lgr_forget(lgr);
 
 	write_lock_bh(&lgr->conns_lock);
 	node = rb_first(&lgr->conns_all);
@@ -400,7 +403,7 @@ static int smc_link_determine_gid(struct smc_link_group *lgr)
 }
 
 /* create a new SMC connection (and a new link group if necessary) */
-int smc_conn_create(struct smc_sock *smc, __be32 peer_in_addr,
+int smc_conn_create(struct smc_sock *smc,
 		    struct smc_ib_device *smcibdev, u8 ibport,
 		    struct smc_clc_msg_local *lcl, int srv_first_contact)
 {
@@ -457,7 +460,7 @@ int smc_conn_create(struct smc_sock *smc, __be32 peer_in_addr,
 
 create:
 	if (local_contact == SMC_FIRST_CONTACT) {
-		rc = smc_lgr_create(smc, peer_in_addr, smcibdev, ibport,
+		rc = smc_lgr_create(smc, smcibdev, ibport,
 				    lcl->id_for_peer, vlan_id);
 		if (rc)
 			goto out;
@@ -698,27 +701,55 @@ static inline int smc_rmb_reserve_rtoken_idx(struct smc_link_group *lgr)
 	return -ENOSPC;
 }
 
-/* save rkey and dma_addr received from peer during clc handshake */
-int smc_rmb_rtoken_handling(struct smc_connection *conn,
-			    struct smc_clc_msg_accept_confirm *clc)
+/* add a new rtoken from peer */
+int smc_rtoken_add(struct smc_link_group *lgr, __be64 nw_vaddr, __be32 nw_rkey)
 {
-	u64 dma_addr = be64_to_cpu(clc->rmb_dma_addr);
-	struct smc_link_group *lgr = conn->lgr;
-	u32 rkey = ntohl(clc->rmb_rkey);
+	u64 dma_addr = be64_to_cpu(nw_vaddr);
+	u32 rkey = ntohl(nw_rkey);
 	int i;
 
 	for (i = 0; i < SMC_RMBS_PER_LGR_MAX; i++) {
 		if ((lgr->rtokens[i][SMC_SINGLE_LINK].rkey == rkey) &&
 		    (lgr->rtokens[i][SMC_SINGLE_LINK].dma_addr == dma_addr) &&
 		    test_bit(i, lgr->rtokens_used_mask)) {
-			conn->rtoken_idx = i;
+			/* already in list */
+			return i;
+		}
+	}
+	i = smc_rmb_reserve_rtoken_idx(lgr);
+	if (i < 0)
+		return i;
+	lgr->rtokens[i][SMC_SINGLE_LINK].rkey = rkey;
+	lgr->rtokens[i][SMC_SINGLE_LINK].dma_addr = dma_addr;
+	return i;
+}
+
+/* delete an rtoken */
+int smc_rtoken_delete(struct smc_link_group *lgr, __be32 nw_rkey)
+{
+	u32 rkey = ntohl(nw_rkey);
+	int i;
+
+	for (i = 0; i < SMC_RMBS_PER_LGR_MAX; i++) {
+		if (lgr->rtokens[i][SMC_SINGLE_LINK].rkey == rkey &&
+		    test_bit(i, lgr->rtokens_used_mask)) {
+			lgr->rtokens[i][SMC_SINGLE_LINK].rkey = 0;
+			lgr->rtokens[i][SMC_SINGLE_LINK].dma_addr = 0;
+
+			clear_bit(i, lgr->rtokens_used_mask);
 			return 0;
 		}
 	}
-	conn->rtoken_idx = smc_rmb_reserve_rtoken_idx(lgr);
+	return -ENOENT;
+}
+
+/* save rkey and dma_addr received from peer during clc handshake */
+int smc_rmb_rtoken_handling(struct smc_connection *conn,
+			    struct smc_clc_msg_accept_confirm *clc)
+{
+	conn->rtoken_idx = smc_rtoken_add(conn->lgr, clc->rmb_dma_addr,
+					  clc->rmb_rkey);
 	if (conn->rtoken_idx < 0)
 		return conn->rtoken_idx;
-	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey = rkey;
-	lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].dma_addr = dma_addr;
 	return 0;
 }

net/smc/smc_core.h

@@ -32,6 +32,12 @@ enum smc_lgr_role {		/* possible roles of a link group */
 	SMC_SERV	/* server */
 };
 
+enum smc_link_state {			/* possible states of a link */
+	SMC_LNK_INACTIVE,	/* link is inactive */
+	SMC_LNK_ACTIVATING,	/* link is being activated */
+	SMC_LNK_ACTIVE		/* link is active */
+};
+
 #define SMC_WR_BUF_SIZE		48	/* size of work request buffer */
 
 struct smc_wr_buf {
@@ -87,8 +93,14 @@ struct smc_link {
 	u8			peer_mac[ETH_ALEN];	/* = gid[8:10||13:15] */
 	u8			peer_gid[sizeof(union ib_gid)];	/* gid of peer*/
 	u8			link_id;	/* unique # within link group */
+
+	enum smc_link_state	state;		/* state of link */
 	struct completion	llc_confirm;	/* wait for rx of conf link */
 	struct completion	llc_confirm_resp; /* wait 4 rx of cnf lnk rsp */
+	int			llc_confirm_rc; /* rc from confirm link msg */
+	int			llc_confirm_resp_rc; /* rc from conf_resp msg */
+	struct completion	llc_add;	/* wait for rx of add link */
+	struct completion	llc_add_resp;	/* wait for rx of add link rsp*/
 };
 
 /* For now we just allow one parallel link per link group. The SMC protocol
@@ -124,7 +136,6 @@ struct smc_rtoken {				/* address/key of remote RMB */
 struct smc_link_group {
 	struct list_head	list;
 	enum smc_lgr_role	role;		/* client or server */
-	__be32			daddr;		/* destination ip address */
 	struct smc_link		lnk[SMC_LINKS_PER_LGR_MAX];	/* smc link */
 	char			peer_systemid[SMC_SYSTEMID_LEN];
 						/* unique system_id of peer */
@@ -186,10 +197,13 @@ struct smc_sock;
 struct smc_clc_msg_accept_confirm;
 
 void smc_lgr_free(struct smc_link_group *lgr);
+void smc_lgr_forget(struct smc_link_group *lgr);
 void smc_lgr_terminate(struct smc_link_group *lgr);
 int smc_buf_create(struct smc_sock *smc);
 int smc_rmb_rtoken_handling(struct smc_connection *conn,
 			    struct smc_clc_msg_accept_confirm *clc);
+int smc_rtoken_add(struct smc_link_group *lgr, __be64 nw_vaddr, __be32 nw_rkey);
+int smc_rtoken_delete(struct smc_link_group *lgr, __be32 nw_rkey);
 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);

net/smc/smc_llc.c

@@ -4,9 +4,6 @@
  *
  *  Link Layer Control (LLC)
  *
- *  For now, we only support the necessary "confirm link" functionality
- *  which happens for the first RoCE link after successful CLC handshake.
- *
  *  Copyright IBM Corp. 2016
  *
  *  Author(s):  Klaus Wacker <Klaus.Wacker@de.ibm.com>
@@ -21,6 +18,122 @@
 #include "smc_clc.h"
 #include "smc_llc.h"
 
+#define SMC_LLC_DATA_LEN		40
+
+struct smc_llc_hdr {
+	struct smc_wr_rx_hdr common;
+	u8 length;	/* 44 */
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u8 reserved:4,
+	   add_link_rej_rsn:4;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u8 add_link_rej_rsn:4,
+	   reserved:4;
+#endif
+	u8 flags;
+};
+
+#define SMC_LLC_FLAG_NO_RMBE_EYEC	0x03
+
+struct smc_llc_msg_confirm_link {	/* type 0x01 */
+	struct smc_llc_hdr hd;
+	u8 sender_mac[ETH_ALEN];
+	u8 sender_gid[SMC_GID_SIZE];
+	u8 sender_qp_num[3];
+	u8 link_num;
+	u8 link_uid[SMC_LGR_ID_SIZE];
+	u8 max_links;
+	u8 reserved[9];
+};
+
+#define SMC_LLC_FLAG_ADD_LNK_REJ	0x40
+#define SMC_LLC_REJ_RSN_NO_ALT_PATH	1
+
+#define SMC_LLC_ADD_LNK_MAX_LINKS	2
+
+struct smc_llc_msg_add_link {		/* type 0x02 */
+	struct smc_llc_hdr hd;
+	u8 sender_mac[ETH_ALEN];
+	u8 reserved2[2];
+	u8 sender_gid[SMC_GID_SIZE];
+	u8 sender_qp_num[3];
+	u8 link_num;
+	u8 flags2;	/* QP mtu */
+	u8 initial_psn[3];
+	u8 reserved[8];
+};
+
+#define SMC_LLC_FLAG_DEL_LINK_ALL	0x40
+#define SMC_LLC_FLAG_DEL_LINK_ORDERLY	0x20
+
+struct smc_llc_msg_del_link {		/* type 0x04 */
+	struct smc_llc_hdr hd;
+	u8 link_num;
+	__be32 reason;
+	u8 reserved[35];
+} __packed;			/* format defined in RFC7609 */
+
+struct smc_llc_msg_test_link {		/* type 0x07 */
+	struct smc_llc_hdr hd;
+	u8 user_data[16];
+	u8 reserved[24];
+};
+
+struct smc_rmb_rtoken {
+	union {
+		u8 num_rkeys;	/* first rtoken byte of CONFIRM LINK msg */
+				/* is actually the num of rtokens, first */
+				/* rtoken is always for the current link */
+		u8 link_id;	/* link id of the rtoken */
+	};
+	__be32 rmb_key;
+	__be64 rmb_vaddr;
+} __packed;			/* format defined in RFC7609 */
+
+#define SMC_LLC_RKEYS_PER_MSG	3
+
+struct smc_llc_msg_confirm_rkey {	/* type 0x06 */
+	struct smc_llc_hdr hd;
+	struct smc_rmb_rtoken rtoken[SMC_LLC_RKEYS_PER_MSG];
+	u8 reserved;
+};
+
+struct smc_llc_msg_confirm_rkey_cont {	/* type 0x08 */
+	struct smc_llc_hdr hd;
+	u8 num_rkeys;
+	struct smc_rmb_rtoken rtoken[SMC_LLC_RKEYS_PER_MSG];
+};
+
+#define SMC_LLC_DEL_RKEY_MAX	8
+#define SMC_LLC_FLAG_RKEY_NEG	0x20
+
+struct smc_llc_msg_delete_rkey {	/* type 0x09 */
+	struct smc_llc_hdr hd;
+	u8 num_rkeys;
+	u8 err_mask;
+	u8 reserved[2];
+	__be32 rkey[8];
+	u8 reserved2[4];
+};
+
+union smc_llc_msg {
+	struct smc_llc_msg_confirm_link confirm_link;
+	struct smc_llc_msg_add_link add_link;
+	struct smc_llc_msg_del_link delete_link;
+
+	struct smc_llc_msg_confirm_rkey confirm_rkey;
+	struct smc_llc_msg_confirm_rkey_cont confirm_rkey_cont;
+	struct smc_llc_msg_delete_rkey delete_rkey;
+
+	struct smc_llc_msg_test_link test_link;
+	struct {
+		struct smc_llc_hdr hdr;
+		u8 data[SMC_LLC_DATA_LEN];
+	} raw;
+};
+
+#define SMC_LLC_FLAG_RESP		0x80
+
 /********************************** send *************************************/
 
 struct smc_llc_tx_pend {
@@ -87,6 +200,7 @@ int smc_llc_send_confirm_link(struct smc_link *link, u8 mac[],
 	memset(confllc, 0, sizeof(*confllc));
 	confllc->hd.common.type = SMC_LLC_CONFIRM_LINK;
 	confllc->hd.length = sizeof(struct smc_llc_msg_confirm_link);
+	confllc->hd.flags |= SMC_LLC_FLAG_NO_RMBE_EYEC;
 	if (reqresp == SMC_LLC_RESP)
 		confllc->hd.flags |= SMC_LLC_FLAG_RESP;
 	memcpy(confllc->sender_mac, mac, ETH_ALEN);
@@ -94,7 +208,104 @@ int smc_llc_send_confirm_link(struct smc_link *link, u8 mac[],
 	hton24(confllc->sender_qp_num, link->roce_qp->qp_num);
 	/* confllc->link_num = SMC_SINGLE_LINK; already done by memset above */
 	memcpy(confllc->link_uid, lgr->id, SMC_LGR_ID_SIZE);
-	confllc->max_links = SMC_LINKS_PER_LGR_MAX;
+	confllc->max_links = SMC_LLC_ADD_LNK_MAX_LINKS; /* enforce peer resp. */
+	/* send llc message */
+	rc = smc_wr_tx_send(link, pend);
+	return rc;
+}
+
+/* send ADD LINK request or response */
+int smc_llc_send_add_link(struct smc_link *link, u8 mac[],
+			  union ib_gid *gid,
+			  enum smc_llc_reqresp reqresp)
+{
+	struct smc_llc_msg_add_link *addllc;
+	struct smc_wr_tx_pend_priv *pend;
+	struct smc_wr_buf *wr_buf;
+	int rc;
+
+	rc = smc_llc_add_pending_send(link, &wr_buf, &pend);
+	if (rc)
+		return rc;
+	addllc = (struct smc_llc_msg_add_link *)wr_buf;
+	memset(addllc, 0, sizeof(*addllc));
+	addllc->hd.common.type = SMC_LLC_ADD_LINK;
+	addllc->hd.length = sizeof(struct smc_llc_msg_add_link);
+	if (reqresp == SMC_LLC_RESP) {
+		addllc->hd.flags |= SMC_LLC_FLAG_RESP;
+		/* always reject more links for now */
+		addllc->hd.flags |= SMC_LLC_FLAG_ADD_LNK_REJ;
+		addllc->hd.add_link_rej_rsn = SMC_LLC_REJ_RSN_NO_ALT_PATH;
+	}
+	memcpy(addllc->sender_mac, mac, ETH_ALEN);
+	memcpy(addllc->sender_gid, gid, SMC_GID_SIZE);
+	/* send llc message */
+	rc = smc_wr_tx_send(link, pend);
+	return rc;
+}
+
+/* send DELETE LINK request or response */
+int smc_llc_send_delete_link(struct smc_link *link,
+			     enum smc_llc_reqresp reqresp)
+{
+	struct smc_llc_msg_del_link *delllc;
+	struct smc_wr_tx_pend_priv *pend;
+	struct smc_wr_buf *wr_buf;
+	int rc;
+
+	rc = smc_llc_add_pending_send(link, &wr_buf, &pend);
+	if (rc)
+		return rc;
+	delllc = (struct smc_llc_msg_del_link *)wr_buf;
+	memset(delllc, 0, sizeof(*delllc));
+	delllc->hd.common.type = SMC_LLC_DELETE_LINK;
+	delllc->hd.length = sizeof(struct smc_llc_msg_add_link);
+	if (reqresp == SMC_LLC_RESP)
+		delllc->hd.flags |= SMC_LLC_FLAG_RESP;
+	/* DEL_LINK_ALL because only 1 link supported */
+	delllc->hd.flags |= SMC_LLC_FLAG_DEL_LINK_ALL;
+	delllc->hd.flags |= SMC_LLC_FLAG_DEL_LINK_ORDERLY;
+	delllc->link_num = link->link_id;
+	/* send llc message */
+	rc = smc_wr_tx_send(link, pend);
+	return rc;
+}
+
+/* send LLC test link request or response */
+int smc_llc_send_test_link(struct smc_link *link, u8 user_data[16],
+			   enum smc_llc_reqresp reqresp)
+{
+	struct smc_llc_msg_test_link *testllc;
+	struct smc_wr_tx_pend_priv *pend;
+	struct smc_wr_buf *wr_buf;
+	int rc;
+
+	rc = smc_llc_add_pending_send(link, &wr_buf, &pend);
+	if (rc)
+		return rc;
+	testllc = (struct smc_llc_msg_test_link *)wr_buf;
+	memset(testllc, 0, sizeof(*testllc));
+	testllc->hd.common.type = SMC_LLC_TEST_LINK;
+	testllc->hd.length = sizeof(struct smc_llc_msg_test_link);
+	if (reqresp == SMC_LLC_RESP)
+		testllc->hd.flags |= SMC_LLC_FLAG_RESP;
+	memcpy(testllc->user_data, user_data, sizeof(testllc->user_data));
+	/* send llc message */
+	rc = smc_wr_tx_send(link, pend);
+	return rc;
+}
+
+/* send a prepared message */
+static int smc_llc_send_message(struct smc_link *link, void *llcbuf, int llclen)
+{
+	struct smc_wr_tx_pend_priv *pend;
+	struct smc_wr_buf *wr_buf;
+	int rc;
+
+	rc = smc_llc_add_pending_send(link, &wr_buf, &pend);
+	if (rc)
+		return rc;
+	memcpy(wr_buf, llcbuf, llclen);
 	/* send llc message */
 	rc = smc_wr_tx_send(link, pend);
 	return rc;
@@ -106,19 +317,156 @@ static void smc_llc_rx_confirm_link(struct smc_link *link,
 				    struct smc_llc_msg_confirm_link *llc)
 {
 	struct smc_link_group *lgr;
+	int conf_rc;
 
 	lgr = container_of(link, struct smc_link_group, lnk[SMC_SINGLE_LINK]);
+
+	/* RMBE eyecatchers are not supported */
+	if (llc->hd.flags & SMC_LLC_FLAG_NO_RMBE_EYEC)
+		conf_rc = 0;
+	else
+		conf_rc = ENOTSUPP;
+
 	if (llc->hd.flags & SMC_LLC_FLAG_RESP) {
-		if (lgr->role == SMC_SERV)
+		if (lgr->role == SMC_SERV &&
+		    link->state == SMC_LNK_ACTIVATING) {
+			link->llc_confirm_resp_rc = conf_rc;
 			complete(&link->llc_confirm_resp);
+		}
 	} else {
-		if (lgr->role == SMC_CLNT) {
+		if (lgr->role == SMC_CLNT &&
+		    link->state == SMC_LNK_ACTIVATING) {
+			link->llc_confirm_rc = conf_rc;
 			link->link_id = llc->link_num;
 			complete(&link->llc_confirm);
 		}
 	}
 }
 
+static void smc_llc_rx_add_link(struct smc_link *link,
+				struct smc_llc_msg_add_link *llc)
+{
+	struct smc_link_group *lgr = container_of(link, struct smc_link_group,
+						  lnk[SMC_SINGLE_LINK]);
+
+	if (llc->hd.flags & SMC_LLC_FLAG_RESP) {
+		if (link->state == SMC_LNK_ACTIVATING)
+			complete(&link->llc_add_resp);
+	} else {
+		if (link->state == SMC_LNK_ACTIVATING) {
+			complete(&link->llc_add);
+			return;
+		}
+
+		if (lgr->role == SMC_SERV) {
+			smc_llc_send_add_link(link,
+					link->smcibdev->mac[link->ibport - 1],
+					&link->smcibdev->gid[link->ibport - 1],
+					SMC_LLC_REQ);
+
+		} else {
+			smc_llc_send_add_link(link,
+					link->smcibdev->mac[link->ibport - 1],
+					&link->smcibdev->gid[link->ibport - 1],
+					SMC_LLC_RESP);
+		}
+	}
+}
+
+static void smc_llc_rx_delete_link(struct smc_link *link,
+				   struct smc_llc_msg_del_link *llc)
+{
+	struct smc_link_group *lgr = container_of(link, struct smc_link_group,
+						  lnk[SMC_SINGLE_LINK]);
+
+	if (llc->hd.flags & SMC_LLC_FLAG_RESP) {
+		if (lgr->role == SMC_SERV)
+			smc_lgr_terminate(lgr);
+	} else {
+		if (lgr->role == SMC_SERV) {
+			smc_lgr_forget(lgr);
+			smc_llc_send_delete_link(link, SMC_LLC_REQ);
+		} else {
+			smc_llc_send_delete_link(link, SMC_LLC_RESP);
+			smc_lgr_terminate(lgr);
+		}
+	}
+}
+
+static void smc_llc_rx_test_link(struct smc_link *link,
+				 struct smc_llc_msg_test_link *llc)
+{
+	if (llc->hd.flags & SMC_LLC_FLAG_RESP) {
+		/* unused as long as we don't send this type of msg */
+	} else {
+		smc_llc_send_test_link(link, llc->user_data, SMC_LLC_RESP);
+	}
+}
+
+static void smc_llc_rx_confirm_rkey(struct smc_link *link,
+				    struct smc_llc_msg_confirm_rkey *llc)
+{
+	struct smc_link_group *lgr;
+	int rc;
+
+	lgr = container_of(link, struct smc_link_group, lnk[SMC_SINGLE_LINK]);
+
+	if (llc->hd.flags & SMC_LLC_FLAG_RESP) {
+		/* unused as long as we don't send this type of msg */
+	} else {
+		rc = smc_rtoken_add(lgr,
+				    llc->rtoken[0].rmb_vaddr,
+				    llc->rtoken[0].rmb_key);
+
+		/* ignore rtokens for other links, we have only one link */
+
+		llc->hd.flags |= SMC_LLC_FLAG_RESP;
+		if (rc < 0)
+			llc->hd.flags |= SMC_LLC_FLAG_RKEY_NEG;
+		smc_llc_send_message(link, (void *)llc, sizeof(*llc));
+	}
+}
+
+static void smc_llc_rx_confirm_rkey_cont(struct smc_link *link,
+				      struct smc_llc_msg_confirm_rkey_cont *llc)
+{
+	if (llc->hd.flags & SMC_LLC_FLAG_RESP) {
+		/* unused as long as we don't send this type of msg */
+	} else {
+		/* ignore rtokens for other links, we have only one link */
+		llc->hd.flags |= SMC_LLC_FLAG_RESP;
+		smc_llc_send_message(link, (void *)llc, sizeof(*llc));
+	}
+}
+
+static void smc_llc_rx_delete_rkey(struct smc_link *link,
+				   struct smc_llc_msg_delete_rkey *llc)
+{
+	struct smc_link_group *lgr;
+	u8 err_mask = 0;
+	int i, max;
+
+	lgr = container_of(link, struct smc_link_group, lnk[SMC_SINGLE_LINK]);
+
+	if (llc->hd.flags & SMC_LLC_FLAG_RESP) {
+		/* unused as long as we don't send this type of msg */
+	} else {
+		max = min_t(u8, llc->num_rkeys, SMC_LLC_DEL_RKEY_MAX);
+		for (i = 0; i < max; i++) {
+			if (smc_rtoken_delete(lgr, llc->rkey[i]))
+				err_mask |= 1 << (SMC_LLC_DEL_RKEY_MAX - 1 - i);
+		}
+
+		if (err_mask) {
+			llc->hd.flags |= SMC_LLC_FLAG_RKEY_NEG;
+			llc->err_mask = err_mask;
+		}
+
+		llc->hd.flags |= SMC_LLC_FLAG_RESP;
+		smc_llc_send_message(link, (void *)llc, sizeof(*llc));
+	}
+}
+
 static void smc_llc_rx_handler(struct ib_wc *wc, void *buf)
 {
 	struct smc_link *link = (struct smc_link *)wc->qp->qp_context;
@@ -128,8 +476,30 @@ static void smc_llc_rx_handler(struct ib_wc *wc, void *buf)
 		return; /* short message */
 	if (llc->raw.hdr.length != sizeof(*llc))
 		return; /* invalid message */
-	if (llc->raw.hdr.common.type == SMC_LLC_CONFIRM_LINK)
+
+	switch (llc->raw.hdr.common.type) {
+	case SMC_LLC_TEST_LINK:
+		smc_llc_rx_test_link(link, &llc->test_link);
+		break;
+	case SMC_LLC_CONFIRM_LINK:
 		smc_llc_rx_confirm_link(link, &llc->confirm_link);
+		break;
+	case SMC_LLC_ADD_LINK:
+		smc_llc_rx_add_link(link, &llc->add_link);
+		break;
+	case SMC_LLC_DELETE_LINK:
+		smc_llc_rx_delete_link(link, &llc->delete_link);
+		break;
+	case SMC_LLC_CONFIRM_RKEY:
+		smc_llc_rx_confirm_rkey(link, &llc->confirm_rkey);
+		break;
+	case SMC_LLC_CONFIRM_RKEY_CONT:
+		smc_llc_rx_confirm_rkey_cont(link, &llc->confirm_rkey_cont);
+		break;
+	case SMC_LLC_DELETE_RKEY:
+		smc_llc_rx_delete_rkey(link, &llc->delete_rkey);
+		break;
+	}
 }
 
 /***************************** init, exit, misc ******************************/
@@ -139,6 +509,30 @@ static struct smc_wr_rx_handler smc_llc_rx_handlers[] = {
 		.handler	= smc_llc_rx_handler,
 		.type		= SMC_LLC_CONFIRM_LINK
 	},
+	{
+		.handler	= smc_llc_rx_handler,
+		.type		= SMC_LLC_TEST_LINK
+	},
+	{
+		.handler	= smc_llc_rx_handler,
+		.type		= SMC_LLC_ADD_LINK
+	},
+	{
+		.handler	= smc_llc_rx_handler,
+		.type		= SMC_LLC_DELETE_LINK
+	},
+	{
+		.handler	= smc_llc_rx_handler,
+		.type		= SMC_LLC_CONFIRM_RKEY
+	},
+	{
+		.handler	= smc_llc_rx_handler,
+		.type		= SMC_LLC_CONFIRM_RKEY_CONT
+	},
+	{
+		.handler	= smc_llc_rx_handler,
+		.type		= SMC_LLC_DELETE_RKEY
+	},
 	{
 		.handler	= NULL,
 	}

net/smc/smc_llc.h

@@ -18,6 +18,7 @@
 #define SMC_LLC_FLAG_RESP		0x80
 
 #define SMC_LLC_WAIT_FIRST_TIME		(5 * HZ)
+#define SMC_LLC_WAIT_TIME		(2 * HZ)
 
 enum smc_llc_reqresp {
 	SMC_LLC_REQ,
@@ -26,39 +27,23 @@ enum smc_llc_reqresp {
 
 enum smc_llc_msg_type {
 	SMC_LLC_CONFIRM_LINK		= 0x01,
-};
-
-#define SMC_LLC_DATA_LEN		40
-
-struct smc_llc_hdr {
-	struct smc_wr_rx_hdr common;
-	u8 length;	/* 44 */
-	u8 reserved;
-	u8 flags;
-};
-
-struct smc_llc_msg_confirm_link {	/* type 0x01 */
-	struct smc_llc_hdr hd;
-	u8 sender_mac[ETH_ALEN];
-	u8 sender_gid[SMC_GID_SIZE];
-	u8 sender_qp_num[3];
-	u8 link_num;
-	u8 link_uid[SMC_LGR_ID_SIZE];
-	u8 max_links;
-	u8 reserved[9];
-};
-
-union smc_llc_msg {
-	struct smc_llc_msg_confirm_link confirm_link;
-	struct {
-		struct smc_llc_hdr hdr;
-		u8 data[SMC_LLC_DATA_LEN];
-	} raw;
+	SMC_LLC_ADD_LINK		= 0x02,
+	SMC_LLC_DELETE_LINK		= 0x04,
+	SMC_LLC_CONFIRM_RKEY		= 0x06,
+	SMC_LLC_TEST_LINK		= 0x07,
+	SMC_LLC_CONFIRM_RKEY_CONT	= 0x08,
+	SMC_LLC_DELETE_RKEY		= 0x09,
 };
 
 /* transmit */
 int smc_llc_send_confirm_link(struct smc_link *lnk, u8 mac[], union ib_gid *gid,
 			      enum smc_llc_reqresp reqresp);
+int smc_llc_send_add_link(struct smc_link *link, u8 mac[], union ib_gid *gid,
+			  enum smc_llc_reqresp reqresp);
+int smc_llc_send_delete_link(struct smc_link *link,
+			     enum smc_llc_reqresp reqresp);
+int smc_llc_send_test_link(struct smc_link *lnk, u8 user_data[16],
+			   enum smc_llc_reqresp reqresp);
 int smc_llc_init(void) __init;
 
 #endif /* SMC_LLC_H */