Merge branch 'net-smc-cleanups'

Ursula Braun says:

====================
net/smc: cleanups 2018-05-18

here are SMC patches for net-next providing restructuring and cleanup
in different areas.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net/smc/af_smc.c

@@ -46,11 +46,6 @@ static DEFINE_MUTEX(smc_create_lgr_pending);	/* serialize link group
 						 * creation
 						 */
 
-struct smc_lgr_list smc_lgr_list = {		/* established link groups */
-	.lock = __SPIN_LOCK_UNLOCKED(smc_lgr_list.lock),
-	.list = LIST_HEAD_INIT(smc_lgr_list.list),
-};
-
 static void smc_tcp_listen_work(struct work_struct *);
 
 static void smc_set_keepalive(struct sock *sk, int val)
@@ -382,10 +377,13 @@ static int smc_clnt_conf_first_link(struct smc_sock *smc)
 static void smc_conn_save_peer_info(struct smc_sock *smc,
 				    struct smc_clc_msg_accept_confirm *clc)
 {
-	smc->conn.peer_conn_idx = clc->conn_idx;
+	int bufsize = smc_uncompress_bufsize(clc->rmbe_size);
+
+	smc->conn.peer_rmbe_idx = clc->rmbe_idx;
 	smc->conn.local_tx_ctrl.token = ntohl(clc->rmbe_alert_token);
-	smc->conn.peer_rmbe_size = smc_uncompress_bufsize(clc->rmbe_size);
+	smc->conn.peer_rmbe_size = bufsize;
 	atomic_set(&smc->conn.peer_rmbe_space, smc->conn.peer_rmbe_size);
+	smc->conn.tx_off = bufsize * (smc->conn.peer_rmbe_idx - 1);
 }
 
 static void smc_link_save_peer_info(struct smc_link *link,
@@ -398,165 +396,186 @@ static void smc_link_save_peer_info(struct smc_link *link,
 	link->peer_mtu = clc->qp_mtu;
 }
 
+/* fall back during connect */
+static int smc_connect_fallback(struct smc_sock *smc)
+{
+	smc->use_fallback = true;
+	smc_copy_sock_settings_to_clc(smc);
+	if (smc->sk.sk_state == SMC_INIT)
+		smc->sk.sk_state = SMC_ACTIVE;
+	return 0;
+}
+
+/* decline and fall back during connect */
+static int smc_connect_decline_fallback(struct smc_sock *smc, int reason_code)
+{
+	int rc;
+
+	if (reason_code < 0) /* error, fallback is not possible */
+		return reason_code;
+	if (reason_code != SMC_CLC_DECL_REPLY) {
+		rc = smc_clc_send_decline(smc, reason_code);
+		if (rc < 0)
+			return rc;
+	}
+	return smc_connect_fallback(smc);
+}
+
+/* abort connecting */
+static int smc_connect_abort(struct smc_sock *smc, int reason_code,
+			     int local_contact)
+{
+	if (local_contact == SMC_FIRST_CONTACT)
+		smc_lgr_forget(smc->conn.lgr);
+	mutex_unlock(&smc_create_lgr_pending);
+	smc_conn_free(&smc->conn);
+	if (reason_code < 0 && smc->sk.sk_state == SMC_INIT)
+		sock_put(&smc->sk); /* passive closing */
+	return reason_code;
+}
+
+/* check if there is a rdma device available for this connection. */
+/* called for connect and listen */
+static int smc_check_rdma(struct smc_sock *smc, struct smc_ib_device **ibdev,
+			  u8 *ibport)
+{
+	int reason_code = 0;
+
+	/* PNET table look up: search active ib_device and port
+	 * within same PNETID that also contains the ethernet device
+	 * used for the internal TCP socket
+	 */
+	smc_pnet_find_roce_resource(smc->clcsock->sk, ibdev, ibport);
+	if (!(*ibdev))
+		reason_code = SMC_CLC_DECL_CNFERR; /* configuration error */
+
+	return reason_code;
+}
+
+/* CLC handshake during connect */
+static int smc_connect_clc(struct smc_sock *smc,
+			   struct smc_clc_msg_accept_confirm *aclc,
+			   struct smc_ib_device *ibdev, u8 ibport)
+{
+	int rc = 0;
+
+	/* do inband token exchange */
+	rc = smc_clc_send_proposal(smc, ibdev, ibport);
+	if (rc)
+		return rc;
+	/* receive SMC Accept CLC message */
+	return smc_clc_wait_msg(smc, aclc, sizeof(*aclc), SMC_CLC_ACCEPT);
+}
+
 /* setup for RDMA connection of client */
-static int smc_connect_rdma(struct smc_sock *smc)
+static int smc_connect_rdma(struct smc_sock *smc,
+			    struct smc_clc_msg_accept_confirm *aclc,
+			    struct smc_ib_device *ibdev, u8 ibport)
+{
+	int local_contact = SMC_FIRST_CONTACT;
+	struct smc_link *link;
+	int reason_code = 0;
+
+	mutex_lock(&smc_create_lgr_pending);
+	local_contact = smc_conn_create(smc, ibdev, ibport, &aclc->lcl,
+					aclc->hdr.flag);
+	if (local_contact < 0) {
+		if (local_contact == -ENOMEM)
+			reason_code = SMC_CLC_DECL_MEM;/* insufficient memory*/
+		else if (local_contact == -ENOLINK)
+			reason_code = SMC_CLC_DECL_SYNCERR; /* synchr. error */
+		else
+			reason_code = SMC_CLC_DECL_INTERR; /* other error */
+		return smc_connect_abort(smc, reason_code, 0);
+	}
+	link = &smc->conn.lgr->lnk[SMC_SINGLE_LINK];
+
+	smc_conn_save_peer_info(smc, aclc);
+
+	/* create send buffer and rmb */
+	if (smc_buf_create(smc))
+		return smc_connect_abort(smc, SMC_CLC_DECL_MEM, local_contact);
+
+	if (local_contact == SMC_FIRST_CONTACT)
+		smc_link_save_peer_info(link, aclc);
+
+	if (smc_rmb_rtoken_handling(&smc->conn, aclc))
+		return smc_connect_abort(smc, SMC_CLC_DECL_INTERR,
+					 local_contact);
+
+	smc_close_init(smc);
+	smc_rx_init(smc);
+
+	if (local_contact == SMC_FIRST_CONTACT) {
+		if (smc_ib_ready_link(link))
+			return smc_connect_abort(smc, SMC_CLC_DECL_INTERR,
+						 local_contact);
+	} else {
+		if (!smc->conn.rmb_desc->reused &&
+		    smc_reg_rmb(link, smc->conn.rmb_desc, true))
+			return smc_connect_abort(smc, SMC_CLC_DECL_INTERR,
+						 local_contact);
+	}
+	smc_rmb_sync_sg_for_device(&smc->conn);
+
+	reason_code = smc_clc_send_confirm(smc);
+	if (reason_code)
+		return smc_connect_abort(smc, reason_code, local_contact);
+
+	smc_tx_init(smc);
+
+	if (local_contact == SMC_FIRST_CONTACT) {
+		/* QP confirmation over RoCE fabric */
+		reason_code = smc_clnt_conf_first_link(smc);
+		if (reason_code)
+			return smc_connect_abort(smc, reason_code,
+						 local_contact);
+	}
+	mutex_unlock(&smc_create_lgr_pending);
+
+	smc_copy_sock_settings_to_clc(smc);
+	if (smc->sk.sk_state == SMC_INIT)
+		smc->sk.sk_state = SMC_ACTIVE;
+
+	return 0;
+}
+
+/* perform steps before actually connecting */
+static int __smc_connect(struct smc_sock *smc)
 {
 	struct smc_clc_msg_accept_confirm aclc;
-	int local_contact = SMC_FIRST_CONTACT;
+	struct smc_ib_device *ibdev;
-	struct smc_ib_device *smcibdev;
-	struct smc_link *link;
-	u8 srv_first_contact;
-	int reason_code = 0;
 	int rc = 0;
 	u8 ibport;
 
 	sock_hold(&smc->sk); /* sock put in passive closing */
 
 	if (smc->use_fallback)
-		goto out_connected;
+		return smc_connect_fallback(smc);
 
-	if (!tcp_sk(smc->clcsock->sk)->syn_smc) {
+	/* if peer has not signalled SMC-capability, fall back */
-		/* peer has not signalled SMC-capability */
+	if (!tcp_sk(smc->clcsock->sk)->syn_smc)
-		smc->use_fallback = true;
+		return smc_connect_fallback(smc);
-		goto out_connected;
-	}
 
 	/* IPSec connections opt out of SMC-R optimizations */
-	if (using_ipsec(smc)) {
+	if (using_ipsec(smc))
-		reason_code = SMC_CLC_DECL_IPSEC;
+		return smc_connect_decline_fallback(smc, SMC_CLC_DECL_IPSEC);
-		goto decline_rdma;
-	}
 
-	/* PNET table look up: search active ib_device and port
+	/* check if a RDMA device is available; if not, fall back */
-	 * within same PNETID that also contains the ethernet device
+	if (smc_check_rdma(smc, &ibdev, &ibport))
-	 * used for the internal TCP socket
+		return smc_connect_decline_fallback(smc, SMC_CLC_DECL_CNFERR);
-	 */
-	smc_pnet_find_roce_resource(smc->clcsock->sk, &smcibdev, &ibport);
-	if (!smcibdev) {
-		reason_code = SMC_CLC_DECL_CNFERR; /* configuration error */
-		goto decline_rdma;
-	}
 
-	/* do inband token exchange */
+	/* perform CLC handshake */
-	reason_code = smc_clc_send_proposal(smc, smcibdev, ibport);
+	rc = smc_connect_clc(smc, &aclc, ibdev, ibport);
-	if (reason_code < 0) {
-		rc = reason_code;
-		goto out_err;
-	}
-	if (reason_code > 0) /* configuration error */
-		goto decline_rdma;
-	/* receive SMC Accept CLC message */
-	reason_code = smc_clc_wait_msg(smc, &aclc, sizeof(aclc),
-				       SMC_CLC_ACCEPT);
-	if (reason_code < 0) {
-		rc = reason_code;
-		goto out_err;
-	}
-	if (reason_code > 0)
-		goto decline_rdma;
-
-	srv_first_contact = aclc.hdr.flag;
-	mutex_lock(&smc_create_lgr_pending);
-	local_contact = smc_conn_create(smc, smcibdev, ibport, &aclc.lcl,
-					srv_first_contact);
-	if (local_contact < 0) {
-		rc = local_contact;
-		if (rc == -ENOMEM)
-			reason_code = SMC_CLC_DECL_MEM;/* insufficient memory*/
-		else if (rc == -ENOLINK)
-			reason_code = SMC_CLC_DECL_SYNCERR; /* synchr. error */
-		else
-			reason_code = SMC_CLC_DECL_INTERR; /* other error */
-		goto decline_rdma_unlock;
-	}
-	link = &smc->conn.lgr->lnk[SMC_SINGLE_LINK];
-
-	smc_conn_save_peer_info(smc, &aclc);
-
-	/* create send buffer and rmb */
-	rc = smc_buf_create(smc);
-	if (rc) {
-		reason_code = SMC_CLC_DECL_MEM;
-		goto decline_rdma_unlock;
-	}
-
-	if (local_contact == SMC_FIRST_CONTACT)
-		smc_link_save_peer_info(link, &aclc);
-
-	rc = smc_rmb_rtoken_handling(&smc->conn, &aclc);
-	if (rc) {
-		reason_code = SMC_CLC_DECL_INTERR;
-		goto decline_rdma_unlock;
-	}
-
-	smc_close_init(smc);
-	smc_rx_init(smc);
-
-	if (local_contact == SMC_FIRST_CONTACT) {
-		rc = smc_ib_ready_link(link);
-		if (rc) {
-			reason_code = SMC_CLC_DECL_INTERR;
-			goto decline_rdma_unlock;
-		}
-	} else {
-		if (!smc->conn.rmb_desc->reused) {
-			if (smc_reg_rmb(link, smc->conn.rmb_desc, true)) {
-				reason_code = SMC_CLC_DECL_INTERR;
-				goto decline_rdma_unlock;
-			}
-		}
-	}
-	smc_rmb_sync_sg_for_device(&smc->conn);
-
-	rc = smc_clc_send_confirm(smc);
 	if (rc)
-		goto out_err_unlock;
+		return smc_connect_decline_fallback(smc, rc);
 
-	if (local_contact == SMC_FIRST_CONTACT) {
+	/* connect using rdma */
-		/* QP confirmation over RoCE fabric */
+	rc = smc_connect_rdma(smc, &aclc, ibdev, ibport);
-		reason_code = smc_clnt_conf_first_link(smc);
+	if (rc)
-		if (reason_code < 0) {
+		return smc_connect_decline_fallback(smc, rc);
-			rc = reason_code;
-			goto out_err_unlock;
-		}
-		if (reason_code > 0)
-			goto decline_rdma_unlock;
-	}
 
-	mutex_unlock(&smc_create_lgr_pending);
+	return 0;
-	smc_tx_init(smc);
-
-out_connected:
-	smc_copy_sock_settings_to_clc(smc);
-	if (smc->sk.sk_state == SMC_INIT)
-		smc->sk.sk_state = SMC_ACTIVE;
-
-	return rc ? rc : local_contact;
-
-decline_rdma_unlock:
-	if (local_contact == SMC_FIRST_CONTACT)
-		smc_lgr_forget(smc->conn.lgr);
-	mutex_unlock(&smc_create_lgr_pending);
-	smc_conn_free(&smc->conn);
-decline_rdma:
-	/* RDMA setup failed, switch back to TCP */
-	smc->use_fallback = true;
-	if (reason_code && (reason_code != SMC_CLC_DECL_REPLY)) {
-		rc = smc_clc_send_decline(smc, reason_code);
-		if (rc < 0)
-			goto out_err;
-	}
-	goto out_connected;
-
-out_err_unlock:
-	if (local_contact == SMC_FIRST_CONTACT)
-		smc_lgr_forget(smc->conn.lgr);
-	mutex_unlock(&smc_create_lgr_pending);
-	smc_conn_free(&smc->conn);
-out_err:
-	if (smc->sk.sk_state == SMC_INIT)
-		sock_put(&smc->sk); /* passive closing */
-	return rc;
 }
 
 static int smc_connect(struct socket *sock, struct sockaddr *addr,
@@ -592,8 +611,7 @@ static int smc_connect(struct socket *sock, struct sockaddr *addr,
 	if (rc)
 		goto out;
 
-	/* setup RDMA connection */
+	rc = __smc_connect(smc);
-	rc = smc_connect_rdma(smc);
 	if (rc < 0)
 		goto out;
 	else
@@ -791,145 +809,12 @@ static int smc_serv_conf_first_link(struct smc_sock *smc)
 	return 0;
 }
 
-/* setup for RDMA connection of server */
+/* listen worker: finish */
-static void smc_listen_work(struct work_struct *work)
+static void smc_listen_out(struct smc_sock *new_smc)
 {
-	struct smc_sock *new_smc = container_of(work, struct smc_sock,
-						smc_listen_work);
-	struct smc_clc_msg_proposal_prefix *pclc_prfx;
-	struct socket *newclcsock = new_smc->clcsock;
 	struct smc_sock *lsmc = new_smc->listen_smc;
-	struct smc_clc_msg_accept_confirm cclc;
-	int local_contact = SMC_REUSE_CONTACT;
 	struct sock *newsmcsk = &new_smc->sk;
-	struct smc_clc_msg_proposal *pclc;
-	struct smc_ib_device *smcibdev;
-	u8 buf[SMC_CLC_MAX_LEN];
-	struct smc_link *link;
-	int reason_code = 0;
-	int rc = 0;
-	u8 ibport;
 
-	if (new_smc->use_fallback)
-		goto out_connected;
-
-	/* check if peer is smc capable */
-	if (!tcp_sk(newclcsock->sk)->syn_smc) {
-		new_smc->use_fallback = true;
-		goto out_connected;
-	}
-
-	/* do inband token exchange -
-	 *wait for and receive SMC Proposal CLC message
-	 */
-	reason_code = smc_clc_wait_msg(new_smc, &buf, sizeof(buf),
-				       SMC_CLC_PROPOSAL);
-	if (reason_code < 0)
-		goto out_err;
-	if (reason_code > 0)
-		goto decline_rdma;
-
-	/* IPSec connections opt out of SMC-R optimizations */
-	if (using_ipsec(new_smc)) {
-		reason_code = SMC_CLC_DECL_IPSEC;
-		goto decline_rdma;
-	}
-
-	/* PNET table look up: search active ib_device and port
-	 * within same PNETID that also contains the ethernet device
-	 * used for the internal TCP socket
-	 */
-	smc_pnet_find_roce_resource(newclcsock->sk, &smcibdev, &ibport);
-	if (!smcibdev) {
-		reason_code = SMC_CLC_DECL_CNFERR; /* configuration error */
-		goto decline_rdma;
-	}
-
-	pclc = (struct smc_clc_msg_proposal *)&buf;
-	pclc_prfx = smc_clc_proposal_get_prefix(pclc);
-
-	rc = smc_clc_prfx_match(newclcsock, pclc_prfx);
-	if (rc) {
-		reason_code = SMC_CLC_DECL_CNFERR; /* configuration error */
-		goto decline_rdma;
-	}
-
-	/* allocate connection / link group */
-	mutex_lock(&smc_create_lgr_pending);
-	local_contact = smc_conn_create(new_smc, smcibdev, ibport, &pclc->lcl,
-					0);
-	if (local_contact < 0) {
-		rc = local_contact;
-		if (rc == -ENOMEM)
-			reason_code = SMC_CLC_DECL_MEM;/* insufficient memory*/
-		goto decline_rdma_unlock;
-	}
-	link = &new_smc->conn.lgr->lnk[SMC_SINGLE_LINK];
-
-	/* create send buffer and rmb */
-	rc = smc_buf_create(new_smc);
-	if (rc) {
-		reason_code = SMC_CLC_DECL_MEM;
-		goto decline_rdma_unlock;
-	}
-
-	smc_close_init(new_smc);
-	smc_rx_init(new_smc);
-
-	if (local_contact != SMC_FIRST_CONTACT) {
-		if (!new_smc->conn.rmb_desc->reused) {
-			if (smc_reg_rmb(link, new_smc->conn.rmb_desc, true)) {
-				reason_code = SMC_CLC_DECL_INTERR;
-				goto decline_rdma_unlock;
-			}
-		}
-	}
-	smc_rmb_sync_sg_for_device(&new_smc->conn);
-
-	rc = smc_clc_send_accept(new_smc, local_contact);
-	if (rc)
-		goto out_err_unlock;
-
-	/* receive SMC Confirm CLC message */
-	reason_code = smc_clc_wait_msg(new_smc, &cclc, sizeof(cclc),
-				       SMC_CLC_CONFIRM);
-	if (reason_code < 0)
-		goto out_err_unlock;
-	if (reason_code > 0)
-		goto decline_rdma_unlock;
-	smc_conn_save_peer_info(new_smc, &cclc);
-	if (local_contact == SMC_FIRST_CONTACT)
-		smc_link_save_peer_info(link, &cclc);
-
-	rc = smc_rmb_rtoken_handling(&new_smc->conn, &cclc);
-	if (rc) {
-		reason_code = SMC_CLC_DECL_INTERR;
-		goto decline_rdma_unlock;
-	}
-
-	if (local_contact == SMC_FIRST_CONTACT) {
-		rc = smc_ib_ready_link(link);
-		if (rc) {
-			reason_code = SMC_CLC_DECL_INTERR;
-			goto decline_rdma_unlock;
-		}
-		/* QP confirmation over RoCE fabric */
-		reason_code = smc_serv_conf_first_link(new_smc);
-		if (reason_code < 0)
-			/* peer is not aware of a problem */
-			goto out_err_unlock;
-		if (reason_code > 0)
-			goto decline_rdma_unlock;
-	}
-
-	smc_tx_init(new_smc);
-	mutex_unlock(&smc_create_lgr_pending);
-
-out_connected:
-	sk_refcnt_debug_inc(newsmcsk);
-	if (newsmcsk->sk_state == SMC_INIT)
-		newsmcsk->sk_state = SMC_ACTIVE;
-enqueue:
 	lock_sock_nested(&lsmc->sk, SINGLE_DEPTH_NESTING);
 	if (lsmc->sk.sk_state == SMC_LISTEN) {
 		smc_accept_enqueue(&lsmc->sk, newsmcsk);
@@ -941,32 +826,222 @@ static void smc_listen_work(struct work_struct *work)
 	/* Wake up accept */
 	lsmc->sk.sk_data_ready(&lsmc->sk);
 	sock_put(&lsmc->sk); /* sock_hold in smc_tcp_listen_work */
-	return;
+}
 
-decline_rdma_unlock:
+/* listen worker: finish in state connected */
-	if (local_contact == SMC_FIRST_CONTACT)
+static void smc_listen_out_connected(struct smc_sock *new_smc)
-		smc_lgr_forget(new_smc->conn.lgr);
+{
-	mutex_unlock(&smc_create_lgr_pending);
+	struct sock *newsmcsk = &new_smc->sk;
-decline_rdma:
+
-	/* RDMA setup failed, switch back to TCP */
+	sk_refcnt_debug_inc(newsmcsk);
-	smc_conn_free(&new_smc->conn);
+	if (newsmcsk->sk_state == SMC_INIT)
-	new_smc->use_fallback = true;
+		newsmcsk->sk_state = SMC_ACTIVE;
-	if (reason_code && (reason_code != SMC_CLC_DECL_REPLY)) {
+
-		if (smc_clc_send_decline(new_smc, reason_code) < 0)
+	smc_listen_out(new_smc);
-			goto out_err;
+}
-	}
+
-	goto out_connected;
+/* listen worker: finish in error state */
+static void smc_listen_out_err(struct smc_sock *new_smc)
+{
+	struct sock *newsmcsk = &new_smc->sk;
 
-out_err_unlock:
-	if (local_contact == SMC_FIRST_CONTACT)
-		smc_lgr_forget(new_smc->conn.lgr);
-	mutex_unlock(&smc_create_lgr_pending);
-out_err:
 	if (newsmcsk->sk_state == SMC_INIT)
 		sock_put(&new_smc->sk); /* passive closing */
 	newsmcsk->sk_state = SMC_CLOSED;
 	smc_conn_free(&new_smc->conn);
-	goto enqueue; /* queue new sock with sk_err set */
+
+	smc_listen_out(new_smc);
+}
+
+/* listen worker: decline and fall back if possible */
+static void smc_listen_decline(struct smc_sock *new_smc, int reason_code,
+			       int local_contact)
+{
+	/* RDMA setup failed, switch back to TCP */
+	if (local_contact == SMC_FIRST_CONTACT)
+		smc_lgr_forget(new_smc->conn.lgr);
+	if (reason_code < 0) { /* error, no fallback possible */
+		smc_listen_out_err(new_smc);
+		return;
+	}
+	smc_conn_free(&new_smc->conn);
+	new_smc->use_fallback = true;
+	if (reason_code && reason_code != SMC_CLC_DECL_REPLY) {
+		if (smc_clc_send_decline(new_smc, reason_code) < 0) {
+			smc_listen_out_err(new_smc);
+			return;
+		}
+	}
+	smc_listen_out_connected(new_smc);
+}
+
+/* listen worker: check prefixes */
+static int smc_listen_rdma_check(struct smc_sock *new_smc,
+				 struct smc_clc_msg_proposal *pclc)
+{
+	struct smc_clc_msg_proposal_prefix *pclc_prfx;
+	struct socket *newclcsock = new_smc->clcsock;
+
+	pclc_prfx = smc_clc_proposal_get_prefix(pclc);
+	if (smc_clc_prfx_match(newclcsock, pclc_prfx))
+		return SMC_CLC_DECL_CNFERR;
+
+	return 0;
+}
+
+/* listen worker: initialize connection and buffers */
+static int smc_listen_rdma_init(struct smc_sock *new_smc,
+				struct smc_clc_msg_proposal *pclc,
+				struct smc_ib_device *ibdev, u8 ibport,
+				int *local_contact)
+{
+	/* allocate connection / link group */
+	*local_contact = smc_conn_create(new_smc, ibdev, ibport, &pclc->lcl, 0);
+	if (*local_contact < 0) {
+		if (*local_contact == -ENOMEM)
+			return SMC_CLC_DECL_MEM;/* insufficient memory*/
+		return SMC_CLC_DECL_INTERR; /* other error */
+	}
+
+	/* create send buffer and rmb */
+	if (smc_buf_create(new_smc))
+		return SMC_CLC_DECL_MEM;
+
+	return 0;
+}
+
+/* listen worker: register buffers */
+static int smc_listen_rdma_reg(struct smc_sock *new_smc, int local_contact)
+{
+	struct smc_link *link = &new_smc->conn.lgr->lnk[SMC_SINGLE_LINK];
+
+	if (local_contact != SMC_FIRST_CONTACT) {
+		if (!new_smc->conn.rmb_desc->reused) {
+			if (smc_reg_rmb(link, new_smc->conn.rmb_desc, true))
+				return SMC_CLC_DECL_INTERR;
+		}
+	}
+	smc_rmb_sync_sg_for_device(&new_smc->conn);
+
+	return 0;
+}
+
+/* listen worker: finish RDMA setup */
+static void smc_listen_rdma_finish(struct smc_sock *new_smc,
+				   struct smc_clc_msg_accept_confirm *cclc,
+				   int local_contact)
+{
+	struct smc_link *link = &new_smc->conn.lgr->lnk[SMC_SINGLE_LINK];
+	int reason_code = 0;
+
+	if (local_contact == SMC_FIRST_CONTACT)
+		smc_link_save_peer_info(link, cclc);
+
+	if (smc_rmb_rtoken_handling(&new_smc->conn, cclc)) {
+		reason_code = SMC_CLC_DECL_INTERR;
+		goto decline;
+	}
+
+	if (local_contact == SMC_FIRST_CONTACT) {
+		if (smc_ib_ready_link(link)) {
+			reason_code = SMC_CLC_DECL_INTERR;
+			goto decline;
+		}
+		/* QP confirmation over RoCE fabric */
+		reason_code = smc_serv_conf_first_link(new_smc);
+		if (reason_code)
+			goto decline;
+	}
+	return;
+
+decline:
+	mutex_unlock(&smc_create_lgr_pending);
+	smc_listen_decline(new_smc, reason_code, local_contact);
+}
+
+/* setup for RDMA connection of server */
+static void smc_listen_work(struct work_struct *work)
+{
+	struct smc_sock *new_smc = container_of(work, struct smc_sock,
+						smc_listen_work);
+	struct socket *newclcsock = new_smc->clcsock;
+	struct smc_clc_msg_accept_confirm cclc;
+	struct smc_clc_msg_proposal *pclc;
+	struct smc_ib_device *ibdev;
+	u8 buf[SMC_CLC_MAX_LEN];
+	int local_contact = 0;
+	int reason_code = 0;
+	int rc = 0;
+	u8 ibport;
+
+	if (new_smc->use_fallback) {
+		smc_listen_out_connected(new_smc);
+		return;
+	}
+
+	/* check if peer is smc capable */
+	if (!tcp_sk(newclcsock->sk)->syn_smc) {
+		new_smc->use_fallback = true;
+		smc_listen_out_connected(new_smc);
+		return;
+	}
+
+	/* do inband token exchange -
+	 * wait for and receive SMC Proposal CLC message
+	 */
+	pclc = (struct smc_clc_msg_proposal *)&buf;
+	reason_code = smc_clc_wait_msg(new_smc, pclc, SMC_CLC_MAX_LEN,
+				       SMC_CLC_PROPOSAL);
+	if (reason_code) {
+		smc_listen_decline(new_smc, reason_code, 0);
+		return;
+	}
+
+	/* IPSec connections opt out of SMC-R optimizations */
+	if (using_ipsec(new_smc)) {
+		smc_listen_decline(new_smc, SMC_CLC_DECL_IPSEC, 0);
+		return;
+	}
+
+	mutex_lock(&smc_create_lgr_pending);
+	smc_close_init(new_smc);
+	smc_rx_init(new_smc);
+	smc_tx_init(new_smc);
+
+	/* check if RDMA is available */
+	if (smc_check_rdma(new_smc, &ibdev, &ibport) ||
+	    smc_listen_rdma_check(new_smc, pclc) ||
+	    smc_listen_rdma_init(new_smc, pclc, ibdev, ibport,
+				 &local_contact) ||
+	    smc_listen_rdma_reg(new_smc, local_contact)) {
+		/* SMC not supported, decline */
+		mutex_unlock(&smc_create_lgr_pending);
+		smc_listen_decline(new_smc, SMC_CLC_DECL_CNFERR, local_contact);
+		return;
+	}
+
+	/* send SMC Accept CLC message */
+	rc = smc_clc_send_accept(new_smc, local_contact);
+	if (rc) {
+		mutex_unlock(&smc_create_lgr_pending);
+		smc_listen_decline(new_smc, rc, local_contact);
+		return;
+	}
+
+	/* receive SMC Confirm CLC message */
+	reason_code = smc_clc_wait_msg(new_smc, &cclc, sizeof(cclc),
+				       SMC_CLC_CONFIRM);
+	if (reason_code) {
+		mutex_unlock(&smc_create_lgr_pending);
+		smc_listen_decline(new_smc, reason_code, local_contact);
+		return;
+	}
+
+	/* finish worker */
+	smc_listen_rdma_finish(new_smc, &cclc, local_contact);
+	smc_conn_save_peer_info(new_smc, &cclc);
+	mutex_unlock(&smc_create_lgr_pending);
+	smc_listen_out_connected(new_smc);
 }
 
 static void smc_tcp_listen_work(struct work_struct *work)
@@ -1227,7 +1302,7 @@ static __poll_t smc_poll(struct file *file, struct socket *sock,
 			if (sk->sk_state == SMC_INIT &&
 			    mask & EPOLLOUT &&
 			    smc->clcsock->sk->sk_state != TCP_CLOSE) {
-				rc = smc_connect_rdma(smc);
+				rc = __smc_connect(smc);
 				if (rc < 0)
 					mask |= EPOLLERR;
 				/* success cases including fallback */
@@ -1421,7 +1496,7 @@ static int smc_ioctl(struct socket *sock, unsigned int cmd,
 		/* output queue size (not send + not acked) */
 		if (smc->sk.sk_state == SMC_LISTEN)
 			return -EINVAL;
-		answ = smc->conn.sndbuf_size -
+		answ = smc->conn.sndbuf_desc->len -
 					atomic_read(&smc->conn.sndbuf_space);
 		break;
 	case SIOCOUTQNSD:
@@ -1637,19 +1712,7 @@ static int __init smc_init(void)
 
 static void __exit smc_exit(void)
 {
-	struct smc_link_group *lgr, *lg;
+	smc_core_exit();
-	LIST_HEAD(lgr_freeing_list);
-
-	spin_lock_bh(&smc_lgr_list.lock);
-	if (!list_empty(&smc_lgr_list.list))
-		list_splice_init(&smc_lgr_list.list, &lgr_freeing_list);
-	spin_unlock_bh(&smc_lgr_list.lock);
-	list_for_each_entry_safe(lgr, lg, &lgr_freeing_list, list) {
-		list_del_init(&lgr->list);
-		smc_llc_link_inactive(&lgr->lnk[SMC_SINGLE_LINK]);
-		cancel_delayed_work_sync(&lgr->free_work);
-		smc_lgr_free(lgr); /* free link group */
-	}
 	static_branch_disable(&tcp_have_smc);
 	smc_ib_unregister_client();
 	sock_unregister(PF_SMC);

net/smc/smc.h

@@ -118,7 +118,7 @@ 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 */
+	u8			peer_rmbe_idx;	/* from tcp handshake */
 	int			peer_rmbe_size;	/* size of peer rx buffer */
 	atomic_t		peer_rmbe_space;/* remaining free bytes in peer
 						 * rmbe
@@ -126,9 +126,7 @@ struct smc_connection {
 	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
@@ -153,6 +151,7 @@ struct smc_connection {
 	u16			tx_cdc_seq;	/* sequence # for CDC send */
 	spinlock_t		send_lock;	/* protect wr_sends */
 	struct delayed_work	tx_work;	/* retry of smc_cdc_msg_send */
+	u32			tx_off;		/* base offset in peer rmb */
 
 	struct smc_host_cdc_msg	local_rx_ctrl;	/* filled during event_handl.
 						 * .prod cf. TCP rcv_nxt
@@ -221,41 +220,6 @@ static inline u32 ntoh24(u8 *net)
 	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)
 {
@@ -269,12 +233,6 @@ static inline bool using_ipsec(struct smc_sock *smc)
 }
 #endif
 
-struct smc_clc_msg_local;
-
-void smc_conn_free(struct smc_connection *conn);
-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);
 void smc_close_non_accepted(struct sock *sk);
 

net/smc/smc_cdc.c

@@ -44,13 +44,13 @@ static void smc_cdc_tx_handler(struct smc_wr_tx_pend_priv *pnd_snd,
 	smc = container_of(cdcpend->conn, struct smc_sock, conn);
 	bh_lock_sock(&smc->sk);
 	if (!wc_status) {
-		diff = smc_curs_diff(cdcpend->conn->sndbuf_size,
+		diff = smc_curs_diff(cdcpend->conn->sndbuf_desc->len,
 				     &cdcpend->conn->tx_curs_fin,
 				     &cdcpend->cursor);
 		/* sndbuf_space is decreased in smc_sendmsg */
 		smp_mb__before_atomic();
 		atomic_add(diff, &cdcpend->conn->sndbuf_space);
-		/* guarantee 0 <= sndbuf_space <= sndbuf_size */
+		/* guarantee 0 <= sndbuf_space <= sndbuf_desc->len */
 		smp_mb__after_atomic();
 		smc_curs_write(&cdcpend->conn->tx_curs_fin,
 			       smc_curs_read(&cdcpend->cursor, cdcpend->conn),
@@ -165,19 +165,12 @@ static inline bool smc_cdc_before(u16 seq1, u16 seq2)
 }
 
 static void smc_cdc_msg_recv_action(struct smc_sock *smc,
-				    struct smc_link *link,
 				    struct smc_cdc_msg *cdc)
 {
 	union smc_host_cursor cons_old, prod_old;
 	struct smc_connection *conn = &smc->conn;
 	int diff_cons, diff_prod;
 
-	if (!cdc->prod_flags.failover_validation) {
-		if (smc_cdc_before(ntohs(cdc->seqno),
-				   conn->local_rx_ctrl.seqno))
-			/* received seqno is old */
-			return;
-	}
 	smc_curs_write(&prod_old,
 		       smc_curs_read(&conn->local_rx_ctrl.prod, conn),
 		       conn);
@@ -198,13 +191,13 @@ static void smc_cdc_msg_recv_action(struct smc_sock *smc,
 		smp_mb__after_atomic();
 	}
 
-	diff_prod = smc_curs_diff(conn->rmbe_size, &prod_old,
+	diff_prod = smc_curs_diff(conn->rmb_desc->len, &prod_old,
 				  &conn->local_rx_ctrl.prod);
 	if (diff_prod) {
 		/* bytes_to_rcv is decreased in smc_recvmsg */
 		smp_mb__before_atomic();
 		atomic_add(diff_prod, &conn->bytes_to_rcv);
-		/* guarantee 0 <= bytes_to_rcv <= rmbe_size */
+		/* guarantee 0 <= bytes_to_rcv <= rmb_desc->len */
 		smp_mb__after_atomic();
 		smc->sk.sk_data_ready(&smc->sk);
 	} else if ((conn->local_rx_ctrl.prod_flags.write_blocked) ||
@@ -236,26 +229,11 @@ static void smc_cdc_msg_recv_action(struct smc_sock *smc,
 }
 
 /* called under tasklet context */
-static inline void smc_cdc_msg_recv(struct smc_cdc_msg *cdc,
+static void smc_cdc_msg_recv(struct smc_sock *smc, struct smc_cdc_msg *cdc)
-				    struct smc_link *link, u64 wr_id)
 {
-	struct smc_link_group *lgr = container_of(link, struct smc_link_group,
-						  lnk[SMC_SINGLE_LINK]);
-	struct smc_connection *connection;
-	struct smc_sock *smc;
-
-	/* lookup connection */
-	read_lock_bh(&lgr->conns_lock);
-	connection = smc_lgr_find_conn(ntohl(cdc->token), lgr);
-	if (!connection) {
-		read_unlock_bh(&lgr->conns_lock);
-		return;
-	}
-	smc = container_of(connection, struct smc_sock, conn);
 	sock_hold(&smc->sk);
-	read_unlock_bh(&lgr->conns_lock);
 	bh_lock_sock(&smc->sk);
-	smc_cdc_msg_recv_action(smc, link, cdc);
+	smc_cdc_msg_recv_action(smc, cdc);
 	bh_unlock_sock(&smc->sk);
 	sock_put(&smc->sk); /* no free sk in softirq-context */
 }
@@ -266,12 +244,31 @@ static void smc_cdc_rx_handler(struct ib_wc *wc, void *buf)
 {
 	struct smc_link *link = (struct smc_link *)wc->qp->qp_context;
 	struct smc_cdc_msg *cdc = buf;
+	struct smc_connection *conn;
+	struct smc_link_group *lgr;
+	struct smc_sock *smc;
 
 	if (wc->byte_len < offsetof(struct smc_cdc_msg, reserved))
 		return; /* short message */
 	if (cdc->len != SMC_WR_TX_SIZE)
 		return; /* invalid message */
-	smc_cdc_msg_recv(cdc, link, wc->wr_id);
+
+	/* lookup connection */
+	lgr = container_of(link, struct smc_link_group, lnk[SMC_SINGLE_LINK]);
+	read_lock_bh(&lgr->conns_lock);
+	conn = smc_lgr_find_conn(ntohl(cdc->token), lgr);
+	read_unlock_bh(&lgr->conns_lock);
+	if (!conn)
+		return;
+	smc = container_of(conn, struct smc_sock, conn);
+
+	if (!cdc->prod_flags.failover_validation) {
+		if (smc_cdc_before(ntohs(cdc->seqno),
+				   conn->local_rx_ctrl.seqno))
+			/* received seqno is old */
+			return;
+	}
+	smc_cdc_msg_recv(smc, cdc);
 }
 
 static struct smc_wr_rx_handler smc_cdc_rx_handlers[] = {

net/smc/smc_clc.c

@@ -442,7 +442,7 @@ int smc_clc_send_confirm(struct smc_sock *smc)
 	hton24(cclc.qpn, link->roce_qp->qp_num);
 	cclc.rmb_rkey =
 		htonl(conn->rmb_desc->mr_rx[SMC_SINGLE_LINK]->rkey);
-	cclc.conn_idx = 1; /* for now: 1 RMB = 1 RMBE */
+	cclc.rmbe_idx = 1; /* for now: 1 RMB = 1 RMBE */
 	cclc.rmbe_alert_token = htonl(conn->alert_token_local);
 	cclc.qp_mtu = min(link->path_mtu, link->peer_mtu);
 	cclc.rmbe_size = conn->rmbe_size_short;
@@ -494,7 +494,7 @@ int smc_clc_send_accept(struct smc_sock *new_smc, int srv_first_contact)
 	hton24(aclc.qpn, link->roce_qp->qp_num);
 	aclc.rmb_rkey =
 		htonl(conn->rmb_desc->mr_rx[SMC_SINGLE_LINK]->rkey);
-	aclc.conn_idx = 1;			/* as long as 1 RMB = 1 RMBE */
+	aclc.rmbe_idx = 1;			/* as long as 1 RMB = 1 RMBE */
 	aclc.rmbe_alert_token = htonl(conn->alert_token_local);
 	aclc.qp_mtu = link->path_mtu;
 	aclc.rmbe_size = conn->rmbe_size_short,

net/smc/smc_clc.h

@@ -97,7 +97,7 @@ struct smc_clc_msg_accept_confirm {	/* clc accept / confirm message */
 	struct smc_clc_msg_local lcl;
 	u8 qpn[3];		/* QP number */
 	__be32 rmb_rkey;	/* RMB rkey */
-	u8 conn_idx;		/* Connection index, which RMBE in RMB */
+	u8 rmbe_idx;		/* Index of RMBE in RMB */
 	__be32 rmbe_alert_token;/* unique connection id */
 #if defined(__BIG_ENDIAN_BITFIELD)
 	u8 rmbe_size : 4,	/* RMBE buf size (compressed notation) */

net/smc/smc_core.c

@@ -30,10 +30,14 @@
 #define SMC_LGR_FREE_DELAY_SERV		(600 * HZ)
 #define SMC_LGR_FREE_DELAY_CLNT		(SMC_LGR_FREE_DELAY_SERV + 10)
 
-static u32 smc_lgr_num;			/* unique link group number */
+static struct smc_lgr_list smc_lgr_list = {	/* established link groups */
+	.lock = __SPIN_LOCK_UNLOCKED(smc_lgr_list.lock),
+	.list = LIST_HEAD_INIT(smc_lgr_list.list),
+	.num = 0,
+};
 
-static void smc_buf_free(struct smc_buf_desc *buf_desc, struct smc_link *lnk,
+static void smc_buf_free(struct smc_link_group *lgr, bool is_rmb,
-			 bool is_rmb);
+			 struct smc_buf_desc *buf_desc);
 
 static void smc_lgr_schedule_free_work(struct smc_link_group *lgr)
 {
@@ -181,8 +185,8 @@ static int smc_lgr_create(struct smc_sock *smc,
 		INIT_LIST_HEAD(&lgr->sndbufs[i]);
 		INIT_LIST_HEAD(&lgr->rmbs[i]);
 	}
-	smc_lgr_num += SMC_LGR_NUM_INCR;
+	smc_lgr_list.num += SMC_LGR_NUM_INCR;
-	memcpy(&lgr->id, (u8 *)&smc_lgr_num, SMC_LGR_ID_SIZE);
+	memcpy(&lgr->id, (u8 *)&smc_lgr_list.num, SMC_LGR_ID_SIZE);
 	INIT_DELAYED_WORK(&lgr->free_work, smc_lgr_free_work);
 	lgr->conns_all = RB_ROOT;
 
@@ -236,26 +240,21 @@ static int smc_lgr_create(struct smc_sock *smc,
 
 static void smc_buf_unuse(struct smc_connection *conn)
 {
-	if (conn->sndbuf_desc) {
+	if (conn->sndbuf_desc)
 		conn->sndbuf_desc->used = 0;
-		conn->sndbuf_size = 0;
-	}
 	if (conn->rmb_desc) {
 		if (!conn->rmb_desc->regerr) {
 			conn->rmb_desc->reused = 1;
 			conn->rmb_desc->used = 0;
-			conn->rmbe_size = 0;
 		} else {
 			/* buf registration failed, reuse not possible */
 			struct smc_link_group *lgr = conn->lgr;
-			struct smc_link *lnk;
 
 			write_lock_bh(&lgr->rmbs_lock);
 			list_del(&conn->rmb_desc->list);
 			write_unlock_bh(&lgr->rmbs_lock);
 
-			lnk = &lgr->lnk[SMC_SINGLE_LINK];
+			smc_buf_free(lgr, true, conn->rmb_desc);
-			smc_buf_free(conn->rmb_desc, lnk, true);
 		}
 	}
 }
@@ -281,9 +280,11 @@ static void smc_link_clear(struct smc_link *lnk)
 	smc_wr_free_link_mem(lnk);
 }
 
-static void smc_buf_free(struct smc_buf_desc *buf_desc, struct smc_link *lnk,
+static void smc_buf_free(struct smc_link_group *lgr, bool is_rmb,
-			 bool is_rmb)
+			 struct smc_buf_desc *buf_desc)
 {
+	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
+
 	if (is_rmb) {
 		if (buf_desc->mr_rx[SMC_SINGLE_LINK])
 			smc_ib_put_memory_region(
@@ -302,7 +303,6 @@ static void smc_buf_free(struct smc_buf_desc *buf_desc, struct smc_link *lnk,
 
 static void __smc_lgr_free_bufs(struct smc_link_group *lgr, bool is_rmb)
 {
-	struct smc_link *lnk = &lgr->lnk[SMC_SINGLE_LINK];
 	struct smc_buf_desc *buf_desc, *bf_desc;
 	struct list_head *buf_list;
 	int i;
@@ -315,7 +315,7 @@ static void __smc_lgr_free_bufs(struct smc_link_group *lgr, bool is_rmb)
 		list_for_each_entry_safe(buf_desc, bf_desc, buf_list,
 					 list) {
 			list_del(&buf_desc->list);
-			smc_buf_free(buf_desc, lnk, is_rmb);
+			smc_buf_free(lgr, is_rmb, buf_desc);
 		}
 	}
 }
@@ -377,6 +377,18 @@ void smc_lgr_terminate(struct smc_link_group *lgr)
 	smc_lgr_schedule_free_work(lgr);
 }
 
+/* Called when IB port is terminated */
+void smc_port_terminate(struct smc_ib_device *smcibdev, u8 ibport)
+{
+	struct smc_link_group *lgr, *l;
+
+	list_for_each_entry_safe(lgr, l, &smc_lgr_list.list, list) {
+		if (lgr->lnk[SMC_SINGLE_LINK].smcibdev == smcibdev &&
+		    lgr->lnk[SMC_SINGLE_LINK].ibport == ibport)
+			smc_lgr_terminate(lgr);
+	}
+}
+
 /* Determine vlan of internal TCP socket.
  * @vlan_id: address to store the determined vlan id into
  */
@@ -461,10 +473,10 @@ int smc_conn_create(struct smc_sock *smc,
 		    struct smc_clc_msg_local *lcl, int srv_first_contact)
 {
 	struct smc_connection *conn = &smc->conn;
+	int local_contact = SMC_FIRST_CONTACT;
 	struct smc_link_group *lgr;
 	unsigned short vlan_id;
 	enum smc_lgr_role role;
-	int local_contact = SMC_FIRST_CONTACT;
 	int rc = 0;
 
 	role = smc->listen_smc ? SMC_SERV : SMC_CLNT;
@@ -530,14 +542,39 @@ int smc_conn_create(struct smc_sock *smc,
 	return rc ? rc : local_contact;
 }
 
+/* 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 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 */
+int smc_uncompress_bufsize(u8 compressed)
+{
+	u32 size;
+
+	size = 0x00000001 << (((int)compressed) + 14);
+	return (int)size;
+}
+
 /* try to reuse a sndbuf or rmb description slot for a certain
  * buffer size; if not available, return NULL
  */
-static inline
+static struct smc_buf_desc *smc_buf_get_slot(int compressed_bufsize,
-struct smc_buf_desc *smc_buf_get_slot(struct smc_link_group *lgr,
+					     rwlock_t *lock,
-				      int compressed_bufsize,
+					     struct list_head *buf_list)
-				      rwlock_t *lock,
-				      struct list_head *buf_list)
 {
 	struct smc_buf_desc *buf_slot;
 
@@ -589,7 +626,7 @@ static struct smc_buf_desc *smc_new_buf_create(struct smc_link_group *lgr,
 	rc = sg_alloc_table(&buf_desc->sgt[SMC_SINGLE_LINK], 1,
 			    GFP_KERNEL);
 	if (rc) {
-		smc_buf_free(buf_desc, lnk, is_rmb);
+		smc_buf_free(lgr, is_rmb, buf_desc);
 		return ERR_PTR(rc);
 	}
 	sg_set_buf(buf_desc->sgt[SMC_SINGLE_LINK].sgl,
@@ -600,7 +637,7 @@ static struct smc_buf_desc *smc_new_buf_create(struct smc_link_group *lgr,
 			       is_rmb ? DMA_FROM_DEVICE : DMA_TO_DEVICE);
 	/* SMC protocol depends on mapping to one DMA address only */
 	if (rc != 1)  {
-		smc_buf_free(buf_desc, lnk, is_rmb);
+		smc_buf_free(lgr, is_rmb, buf_desc);
 		return ERR_PTR(-EAGAIN);
 	}
 
@@ -611,19 +648,20 @@ static struct smc_buf_desc *smc_new_buf_create(struct smc_link_group *lgr,
 					      IB_ACCESS_LOCAL_WRITE,
 					      buf_desc);
 		if (rc) {
-			smc_buf_free(buf_desc, lnk, is_rmb);
+			smc_buf_free(lgr, is_rmb, buf_desc);
 			return ERR_PTR(rc);
 		}
 	}
 
+	buf_desc->len = bufsize;
 	return buf_desc;
 }
 
 static int __smc_buf_create(struct smc_sock *smc, bool is_rmb)
 {
+	struct smc_buf_desc *buf_desc = ERR_PTR(-ENOMEM);
 	struct smc_connection *conn = &smc->conn;
 	struct smc_link_group *lgr = conn->lgr;
-	struct smc_buf_desc *buf_desc = ERR_PTR(-ENOMEM);
 	struct list_head *buf_list;
 	int bufsize, bufsize_short;
 	int sk_buf_size;
@@ -651,7 +689,7 @@ static int __smc_buf_create(struct smc_sock *smc, bool is_rmb)
 			continue;
 
 		/* check for reusable slot in the link group */
-		buf_desc = smc_buf_get_slot(lgr, bufsize_short, lock, buf_list);
+		buf_desc = smc_buf_get_slot(bufsize_short, lock, buf_list);
 		if (buf_desc) {
 			memset(buf_desc->cpu_addr, 0, bufsize);
 			break; /* found reusable slot */
@@ -675,14 +713,12 @@ static int __smc_buf_create(struct smc_sock *smc, bool is_rmb)
 
 	if (is_rmb) {
 		conn->rmb_desc = buf_desc;
-		conn->rmbe_size = bufsize;
 		conn->rmbe_size_short = bufsize_short;
 		smc->sk.sk_rcvbuf = bufsize * 2;
 		atomic_set(&conn->bytes_to_rcv, 0);
 		conn->rmbe_update_limit = smc_rmb_wnd_update_limit(bufsize);
 	} else {
 		conn->sndbuf_desc = buf_desc;
-		conn->sndbuf_size = bufsize;
 		smc->sk.sk_sndbuf = bufsize * 2;
 		atomic_set(&conn->sndbuf_space, bufsize);
 	}
@@ -738,8 +774,7 @@ int smc_buf_create(struct smc_sock *smc)
 	/* create rmb */
 	rc = __smc_buf_create(smc, true);
 	if (rc)
-		smc_buf_free(smc->conn.sndbuf_desc,
+		smc_buf_free(smc->conn.lgr, false, smc->conn.sndbuf_desc);
-			     &smc->conn.lgr->lnk[SMC_SINGLE_LINK], false);
 	return rc;
 }
 
@@ -806,3 +841,21 @@ int smc_rmb_rtoken_handling(struct smc_connection *conn,
 		return conn->rtoken_idx;
 	return 0;
 }
+
+/* Called (from smc_exit) when module is removed */
+void smc_core_exit(void)
+{
+	struct smc_link_group *lgr, *lg;
+	LIST_HEAD(lgr_freeing_list);
+
+	spin_lock_bh(&smc_lgr_list.lock);
+	if (!list_empty(&smc_lgr_list.list))
+		list_splice_init(&smc_lgr_list.list, &lgr_freeing_list);
+	spin_unlock_bh(&smc_lgr_list.lock);
+	list_for_each_entry_safe(lgr, lg, &lgr_freeing_list, list) {
+		list_del_init(&lgr->list);
+		smc_llc_link_inactive(&lgr->lnk[SMC_SINGLE_LINK]);
+		cancel_delayed_work_sync(&lgr->free_work);
+		smc_lgr_free(lgr); /* free link group */
+	}
+}

net/smc/smc_core.h

@@ -23,10 +23,9 @@
 struct smc_lgr_list {			/* list of link group definition */
 	struct list_head	list;
 	spinlock_t		lock;	/* protects list of link groups */
+	u32			num;	/* unique link group number */
 };
 
-extern struct smc_lgr_list	smc_lgr_list; /* list of link groups */
-
 enum smc_lgr_role {		/* possible roles of a link group */
 	SMC_CLNT,	/* client */
 	SMC_SERV	/* server */
@@ -124,6 +123,7 @@ struct smc_buf_desc {
 	struct list_head	list;
 	void			*cpu_addr;	/* virtual address of buffer */
 	struct page		*pages;
+	int			len;		/* length of buffer */
 	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
@@ -141,6 +141,12 @@ struct smc_rtoken {				/* address/key of remote RMB */
 };
 
 #define SMC_LGR_ID_SIZE		4
+#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)
+ */
 
 struct smc_link_group {
 	struct list_head	list;
@@ -205,11 +211,14 @@ static inline struct smc_connection *smc_lgr_find_conn(
 
 struct smc_sock;
 struct smc_clc_msg_accept_confirm;
+struct smc_clc_msg_local;
 
 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);
+void smc_port_terminate(struct smc_ib_device *smcibdev, u8 ibport);
 int smc_buf_create(struct smc_sock *smc);
+int smc_uncompress_bufsize(u8 compressed);
 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);
@@ -218,4 +227,9 @@ 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);
+void smc_conn_free(struct smc_connection *conn);
+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);
+void smc_core_exit(void);
 #endif

net/smc/smc_diag.c

@@ -101,8 +101,9 @@ static int __smc_diag_dump(struct sock *sk, struct sk_buff *skb,
 		struct smc_connection *conn = &smc->conn;
 		struct smc_diag_conninfo cinfo = {
 			.token = conn->alert_token_local,
-			.sndbuf_size = conn->sndbuf_size,
+			.sndbuf_size = conn->sndbuf_desc ?
-			.rmbe_size = conn->rmbe_size,
+				conn->sndbuf_desc->len : 0,
+			.rmbe_size = conn->rmb_desc ? conn->rmb_desc->len : 0,
 			.peer_rmbe_size = conn->peer_rmbe_size,
 
 			.rx_prod.wrap = conn->local_rx_ctrl.prod.wrap,

net/smc/smc_ib.c

@@ -143,17 +143,6 @@ int smc_ib_ready_link(struct smc_link *lnk)
 	return rc;
 }
 
-static void smc_ib_port_terminate(struct smc_ib_device *smcibdev, u8 ibport)
-{
-	struct smc_link_group *lgr, *l;
-
-	list_for_each_entry_safe(lgr, l, &smc_lgr_list.list, list) {
-		if (lgr->lnk[SMC_SINGLE_LINK].smcibdev == smcibdev &&
-		    lgr->lnk[SMC_SINGLE_LINK].ibport == ibport)
-			smc_lgr_terminate(lgr);
-	}
-}
-
 /* process context wrapper for might_sleep smc_ib_remember_port_attr */
 static void smc_ib_port_event_work(struct work_struct *work)
 {
@@ -165,7 +154,7 @@ static void smc_ib_port_event_work(struct work_struct *work)
 		smc_ib_remember_port_attr(smcibdev, port_idx + 1);
 		clear_bit(port_idx, &smcibdev->port_event_mask);
 		if (!smc_ib_port_active(smcibdev, port_idx + 1))
-			smc_ib_port_terminate(smcibdev, port_idx + 1);
+			smc_port_terminate(smcibdev, port_idx + 1);
 	}
 }
 

net/smc/smc_rx.c

@@ -51,7 +51,7 @@ static void smc_rx_wake_up(struct sock *sk)
 static void smc_rx_update_consumer(struct smc_connection *conn,
 				   union smc_host_cursor cons, size_t len)
 {
-	smc_curs_add(conn->rmbe_size, &cons, len);
+	smc_curs_add(conn->rmb_desc->len, &cons, len);
 	smc_curs_write(&conn->local_tx_ctrl.cons, smc_curs_read(&cons, conn),
 		       conn);
 	/* send consumer cursor update if required */
@@ -288,11 +288,11 @@ int smc_rx_recvmsg(struct smc_sock *smc, struct msghdr *msg,
 			       conn);
 		/* subsequent splice() calls pick up where previous left */
 		if (splbytes)
-			smc_curs_add(conn->rmbe_size, &cons, splbytes);
+			smc_curs_add(conn->rmb_desc->len, &cons, splbytes);
 		/* determine chunks where to read from rcvbuf */
 		/* either unwrapped case, or 1st chunk of wrapped case */
-		chunk_len = min_t(size_t,
+		chunk_len = min_t(size_t, copylen, conn->rmb_desc->len -
-				  copylen, conn->rmbe_size - cons.count);
+				  cons.count);
 		chunk_len_sum = chunk_len;
 		chunk_off = cons.count;
 		smc_rmb_sync_sg_for_cpu(conn);
@@ -331,7 +331,7 @@ int smc_rx_recvmsg(struct smc_sock *smc, struct msghdr *msg,
 			/* increased in recv tasklet smc_cdc_msg_rcv() */
 			smp_mb__before_atomic();
 			atomic_sub(copylen, &conn->bytes_to_rcv);
-			/* guarantee 0 <= bytes_to_rcv <= rmbe_size */
+			/* guarantee 0 <= bytes_to_rcv <= rmb_desc->len */
 			smp_mb__after_atomic();
 			if (msg)
 				smc_rx_update_consumer(conn, cons, copylen);

net/smc/smc_tx.c

@@ -180,8 +180,8 @@ int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len)
 		tx_cnt_prep = prep.count;
 		/* determine chunks where to write into sndbuf */
 		/* either unwrapped case, or 1st chunk of wrapped case */
-		chunk_len = min_t(size_t,
+		chunk_len = min_t(size_t, copylen, conn->sndbuf_desc->len -
-				  copylen, conn->sndbuf_size - tx_cnt_prep);
+				  tx_cnt_prep);
 		chunk_len_sum = chunk_len;
 		chunk_off = tx_cnt_prep;
 		smc_sndbuf_sync_sg_for_cpu(conn);
@@ -206,21 +206,21 @@ int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len)
 		}
 		smc_sndbuf_sync_sg_for_device(conn);
 		/* update cursors */
-		smc_curs_add(conn->sndbuf_size, &prep, copylen);
+		smc_curs_add(conn->sndbuf_desc->len, &prep, copylen);
 		smc_curs_write(&conn->tx_curs_prep,
 			       smc_curs_read(&prep, conn),
 			       conn);
 		/* increased in send tasklet smc_cdc_tx_handler() */
 		smp_mb__before_atomic();
 		atomic_sub(copylen, &conn->sndbuf_space);
-		/* guarantee 0 <= sndbuf_space <= sndbuf_size */
+		/* guarantee 0 <= sndbuf_space <= sndbuf_desc->len */
 		smp_mb__after_atomic();
 		/* since we just produced more new data into sndbuf,
 		 * trigger sndbuf consumer: RDMA write into peer RMBE and CDC
 		 */
 		if ((msg->msg_flags & MSG_MORE || smc_tx_is_corked(smc)) &&
 		    (atomic_read(&conn->sndbuf_space) >
-						(conn->sndbuf_size >> 1)))
+						(conn->sndbuf_desc->len >> 1)))
 			/* for a corked socket defer the RDMA writes if there
 			 * is still sufficient sndbuf_space available
 			 */
@@ -261,7 +261,7 @@ static int smc_tx_rdma_write(struct smc_connection *conn, int peer_rmbe_offset,
 	rdma_wr.remote_addr =
 		lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].dma_addr +
 		/* RMBE within RMB */
-		((conn->peer_conn_idx - 1) * conn->peer_rmbe_size) +
+		conn->tx_off +
 		/* offset within RMBE */
 		peer_rmbe_offset;
 	rdma_wr.rkey = lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey;
@@ -286,7 +286,7 @@ static inline void smc_tx_advance_cursors(struct smc_connection *conn,
 	atomic_sub(len, &conn->peer_rmbe_space);
 	/* guarantee 0 <= peer_rmbe_space <= peer_rmbe_size */
 	smp_mb__after_atomic();
-	smc_curs_add(conn->sndbuf_size, sent, len);
+	smc_curs_add(conn->sndbuf_desc->len, sent, len);
 }
 
 /* sndbuf consumer: prepare all necessary (src&dst) chunks of data transmit;
@@ -309,7 +309,7 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
 	smc_curs_write(&sent, smc_curs_read(&conn->tx_curs_sent, conn), conn);
 	smc_curs_write(&prep, smc_curs_read(&conn->tx_curs_prep, conn), conn);
 	/* cf. wmem_alloc - (snd_max - snd_una) */
-	to_send = smc_curs_diff(conn->sndbuf_size, &sent, &prep);
+	to_send = smc_curs_diff(conn->sndbuf_desc->len, &sent, &prep);
 	if (to_send <= 0)
 		return 0;
 
@@ -351,12 +351,12 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
 	dst_len_sum = dst_len;
 	src_off = sent.count;
 	/* dst_len determines the maximum src_len */
-	if (sent.count + dst_len <= conn->sndbuf_size) {
+	if (sent.count + dst_len <= conn->sndbuf_desc->len) {
 		/* unwrapped src case: single chunk of entire dst_len */
 		src_len = dst_len;
 	} else {
 		/* wrapped src case: 2 chunks of sum dst_len; start with 1st: */
-		src_len = conn->sndbuf_size - sent.count;
+		src_len = conn->sndbuf_desc->len - sent.count;
 	}
 	src_len_sum = src_len;
 	dma_addr = sg_dma_address(conn->sndbuf_desc->sgt[SMC_SINGLE_LINK].sgl);
@@ -368,8 +368,8 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
 			sges[srcchunk].lkey = link->roce_pd->local_dma_lkey;
 			num_sges++;
 			src_off += src_len;
-			if (src_off >= conn->sndbuf_size)
+			if (src_off >= conn->sndbuf_desc->len)
-				src_off -= conn->sndbuf_size;
+				src_off -= conn->sndbuf_desc->len;
 						/* modulo in send ring */
 			if (src_len_sum == dst_len)
 				break; /* either on 1st or 2nd iteration */
@@ -387,7 +387,7 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
 		dst_len = len - dst_len; /* remainder */
 		dst_len_sum += dst_len;
 		src_len = min_t(int,
-				dst_len, conn->sndbuf_size - sent.count);
+				dst_len, conn->sndbuf_desc->len - sent.count);
 		src_len_sum = src_len;
 	}
 
@@ -484,11 +484,11 @@ void smc_tx_consumer_update(struct smc_connection *conn)
 	smc_curs_write(&cfed,
 		       smc_curs_read(&conn->rx_curs_confirmed, conn),
 		       conn);
-	to_confirm = smc_curs_diff(conn->rmbe_size, &cfed, &cons);
+	to_confirm = smc_curs_diff(conn->rmb_desc->len, &cfed, &cons);
 
 	if (conn->local_rx_ctrl.prod_flags.cons_curs_upd_req ||
 	    ((to_confirm > conn->rmbe_update_limit) &&
-	     ((to_confirm > (conn->rmbe_size / 2)) ||
+	     ((to_confirm > (conn->rmb_desc->len / 2)) ||
 	      conn->local_rx_ctrl.prod_flags.write_blocked))) {
 		if ((smc_cdc_get_slot_and_msg_send(conn) < 0) &&
 		    conn->alert_token_local) { /* connection healthy */

net/smc/smc_tx.h

@@ -24,7 +24,7 @@ static inline int smc_tx_prepared_sends(struct smc_connection *conn)
 
 	smc_curs_write(&sent, smc_curs_read(&conn->tx_curs_sent, conn), conn);
 	smc_curs_write(&prep, smc_curs_read(&conn->tx_curs_prep, conn), conn);
-	return smc_curs_diff(conn->sndbuf_size, &sent, &prep);
+	return smc_curs_diff(conn->sndbuf_desc->len, &sent, &prep);
 }
 
 void smc_tx_work(struct work_struct *work);