cma.c

linux 内核源代码

	struct ib_cm_sidr_rep_event_param *rep = &ib_event->param.sidr_rep_rcvd;
	int ret = 0;

	if (cma_disable_remove(id_priv, CMA_CONNECT))
		return 0;

	memset(&event, 0, sizeof event);
	switch (ib_event->event) {
	case IB_CM_SIDR_REQ_ERROR:
		event.event = RDMA_CM_EVENT_UNREACHABLE;
		event.status = -ETIMEDOUT;
		break;
	case IB_CM_SIDR_REP_RECEIVED:
		event.param.ud.private_data = ib_event->private_data;
		event.param.ud.private_data_len = IB_CM_SIDR_REP_PRIVATE_DATA_SIZE;
		if (rep->status != IB_SIDR_SUCCESS) {
			event.event = RDMA_CM_EVENT_UNREACHABLE;
			event.status = ib_event->param.sidr_rep_rcvd.status;
			break;
		}
		if (id_priv->qkey != rep->qkey) {
			event.event = RDMA_CM_EVENT_UNREACHABLE;
			event.status = -EINVAL;
			break;
		}
		ib_init_ah_from_path(id_priv->id.device, id_priv->id.port_num,
				     id_priv->id.route.path_rec,
				     &event.param.ud.ah_attr);
		event.param.ud.qp_num = rep->qpn;
		event.param.ud.qkey = rep->qkey;
		event.event = RDMA_CM_EVENT_ESTABLISHED;
		event.status = 0;
		break;
	default:
		printk(KERN_ERR "RDMA CMA: unexpected IB CM event: %d",
		       ib_event->event);
		goto out;
	}

	ret = id_priv->id.event_handler(&id_priv->id, &event);
	if (ret) {
		/* Destroy the CM ID by returning a non-zero value. */
		id_priv->cm_id.ib = NULL;
		cma_exch(id_priv, CMA_DESTROYING);
		cma_enable_remove(id_priv);
		rdma_destroy_id(&id_priv->id);
		return ret;
	}
out:
	cma_enable_remove(id_priv);
	return ret;
}

static int cma_resolve_ib_udp(struct rdma_id_private *id_priv,
			      struct rdma_conn_param *conn_param)
{
	struct ib_cm_sidr_req_param req;
	struct rdma_route *route;
	int ret;

	req.private_data_len = sizeof(struct cma_hdr) +
			       conn_param->private_data_len;
	req.private_data = kzalloc(req.private_data_len, GFP_ATOMIC);
	if (!req.private_data)
		return -ENOMEM;

	if (conn_param->private_data && conn_param->private_data_len)
		memcpy((void *) req.private_data + sizeof(struct cma_hdr),
		       conn_param->private_data, conn_param->private_data_len);

	route = &id_priv->id.route;
	ret = cma_format_hdr((void *) req.private_data, id_priv->id.ps, route);
	if (ret)
		goto out;

	id_priv->cm_id.ib = ib_create_cm_id(id_priv->id.device,
					    cma_sidr_rep_handler, id_priv);
	if (IS_ERR(id_priv->cm_id.ib)) {
		ret = PTR_ERR(id_priv->cm_id.ib);
		goto out;
	}

	req.path = route->path_rec;
	req.service_id = cma_get_service_id(id_priv->id.ps,
					    &route->addr.dst_addr);
	req.timeout_ms = 1 << (CMA_CM_RESPONSE_TIMEOUT - 8);
	req.max_cm_retries = CMA_MAX_CM_RETRIES;

	ret = ib_send_cm_sidr_req(id_priv->cm_id.ib, &req);
	if (ret) {
		ib_destroy_cm_id(id_priv->cm_id.ib);
		id_priv->cm_id.ib = NULL;
	}
out:
	kfree(req.private_data);
	return ret;
}

static int cma_connect_ib(struct rdma_id_private *id_priv,
			  struct rdma_conn_param *conn_param)
{
	struct ib_cm_req_param req;
	struct rdma_route *route;
	void *private_data;
	int offset, ret;

	memset(&req, 0, sizeof req);
	offset = cma_user_data_offset(id_priv->id.ps);
	req.private_data_len = offset + conn_param->private_data_len;
	private_data = kzalloc(req.private_data_len, GFP_ATOMIC);
	if (!private_data)
		return -ENOMEM;

	if (conn_param->private_data && conn_param->private_data_len)
		memcpy(private_data + offset, conn_param->private_data,
		       conn_param->private_data_len);

	id_priv->cm_id.ib = ib_create_cm_id(id_priv->id.device, cma_ib_handler,
					    id_priv);
	if (IS_ERR(id_priv->cm_id.ib)) {
		ret = PTR_ERR(id_priv->cm_id.ib);
		goto out;
	}

	route = &id_priv->id.route;
	ret = cma_format_hdr(private_data, id_priv->id.ps, route);
	if (ret)
		goto out;
	req.private_data = private_data;

	req.primary_path = &route->path_rec[0];
	if (route->num_paths == 2)
		req.alternate_path = &route->path_rec[1];

	req.service_id = cma_get_service_id(id_priv->id.ps,
					    &route->addr.dst_addr);
	req.qp_num = id_priv->qp_num;
	req.qp_type = IB_QPT_RC;
	req.starting_psn = id_priv->seq_num;
	req.responder_resources = conn_param->responder_resources;
	req.initiator_depth = conn_param->initiator_depth;
	req.flow_control = conn_param->flow_control;
	req.retry_count = conn_param->retry_count;
	req.rnr_retry_count = conn_param->rnr_retry_count;
	req.remote_cm_response_timeout = CMA_CM_RESPONSE_TIMEOUT;
	req.local_cm_response_timeout = CMA_CM_RESPONSE_TIMEOUT;
	req.max_cm_retries = CMA_MAX_CM_RETRIES;
	req.srq = id_priv->srq ? 1 : 0;

	ret = ib_send_cm_req(id_priv->cm_id.ib, &req);
out:
	if (ret && !IS_ERR(id_priv->cm_id.ib)) {
		ib_destroy_cm_id(id_priv->cm_id.ib);
		id_priv->cm_id.ib = NULL;
	}

	kfree(private_data);
	return ret;
}

static int cma_connect_iw(struct rdma_id_private *id_priv,
			  struct rdma_conn_param *conn_param)
{
	struct iw_cm_id *cm_id;
	struct sockaddr_in* sin;
	int ret;
	struct iw_cm_conn_param iw_param;

	cm_id = iw_create_cm_id(id_priv->id.device, cma_iw_handler, id_priv);
	if (IS_ERR(cm_id)) {
		ret = PTR_ERR(cm_id);
		goto out;
	}

	id_priv->cm_id.iw = cm_id;

	sin = (struct sockaddr_in*) &id_priv->id.route.addr.src_addr;
	cm_id->local_addr = *sin;

	sin = (struct sockaddr_in*) &id_priv->id.route.addr.dst_addr;
	cm_id->remote_addr = *sin;

	ret = cma_modify_qp_rtr(id_priv);
	if (ret)
		goto out;

	iw_param.ord = conn_param->initiator_depth;
	iw_param.ird = conn_param->responder_resources;
	iw_param.private_data = conn_param->private_data;
	iw_param.private_data_len = conn_param->private_data_len;
	if (id_priv->id.qp)
		iw_param.qpn = id_priv->qp_num;
	else
		iw_param.qpn = conn_param->qp_num;
	ret = iw_cm_connect(cm_id, &iw_param);
out:
	if (ret && !IS_ERR(cm_id)) {
		iw_destroy_cm_id(cm_id);
		id_priv->cm_id.iw = NULL;
	}
	return ret;
}

int rdma_connect(struct rdma_cm_id *id, struct rdma_conn_param *conn_param)
{
	struct rdma_id_private *id_priv;
	int ret;

	id_priv = container_of(id, struct rdma_id_private, id);
	if (!cma_comp_exch(id_priv, CMA_ROUTE_RESOLVED, CMA_CONNECT))
		return -EINVAL;

	if (!id->qp) {
		id_priv->qp_num = conn_param->qp_num;
		id_priv->srq = conn_param->srq;
	}

	switch (rdma_node_get_transport(id->device->node_type)) {
	case RDMA_TRANSPORT_IB:
		if (cma_is_ud_ps(id->ps))
			ret = cma_resolve_ib_udp(id_priv, conn_param);
		else
			ret = cma_connect_ib(id_priv, conn_param);
		break;
	case RDMA_TRANSPORT_IWARP:
		ret = cma_connect_iw(id_priv, conn_param);
		break;
	default:
		ret = -ENOSYS;
		break;
	}
	if (ret)
		goto err;

	return 0;
err:
	cma_comp_exch(id_priv, CMA_CONNECT, CMA_ROUTE_RESOLVED);
	return ret;
}
EXPORT_SYMBOL(rdma_connect);

static int cma_accept_ib(struct rdma_id_private *id_priv,
			 struct rdma_conn_param *conn_param)
{
	struct ib_cm_rep_param rep;
	struct ib_qp_attr qp_attr;
	int qp_attr_mask, ret;

	if (id_priv->id.qp) {
		ret = cma_modify_qp_rtr(id_priv);
		if (ret)
			goto out;

		qp_attr.qp_state = IB_QPS_RTS;
		ret = ib_cm_init_qp_attr(id_priv->cm_id.ib, &qp_attr,
					 &qp_attr_mask);
		if (ret)
			goto out;

		qp_attr.max_rd_atomic = conn_param->initiator_depth;
		ret = ib_modify_qp(id_priv->id.qp, &qp_attr, qp_attr_mask);
		if (ret)
			goto out;
	}

	memset(&rep, 0, sizeof rep);
	rep.qp_num = id_priv->qp_num;
	rep.starting_psn = id_priv->seq_num;
	rep.private_data = conn_param->private_data;
	rep.private_data_len = conn_param->private_data_len;
	rep.responder_resources = conn_param->responder_resources;
	rep.initiator_depth = conn_param->initiator_depth;
	rep.failover_accepted = 0;
	rep.flow_control = conn_param->flow_control;
	rep.rnr_retry_count = conn_param->rnr_retry_count;
	rep.srq = id_priv->srq ? 1 : 0;

	ret = ib_send_cm_rep(id_priv->cm_id.ib, &rep);
out:
	return ret;
}

static int cma_accept_iw(struct rdma_id_private *id_priv,
		  struct rdma_conn_param *conn_param)
{
	struct iw_cm_conn_param iw_param;
	int ret;

	ret = cma_modify_qp_rtr(id_priv);
	if (ret)
		return ret;

	iw_param.ord = conn_param->initiator_depth;
	iw_param.ird = conn_param->responder_resources;
	iw_param.private_data = conn_param->private_data;
	iw_param.private_data_len = conn_param->private_data_len;
	if (id_priv->id.qp) {
		iw_param.qpn = id_priv->qp_num;
	} else
		iw_param.qpn = conn_param->qp_num;

	return iw_cm_accept(id_priv->cm_id.iw, &iw_param);
}

static int cma_send_sidr_rep(struct rdma_id_private *id_priv,
			     enum ib_cm_sidr_status status,
			     const void *private_data, int private_data_len)
{
	struct ib_cm_sidr_rep_param rep;

	memset(&rep, 0, sizeof rep);
	rep.status = status;
	if (status == IB_SIDR_SUCCESS) {
		rep.qp_num = id_priv->qp_num;
		rep.qkey = id_priv->qkey;
	}
	rep.private_data = private_data;
	rep.private_data_len = private_data_len;

	return ib_send_cm_sidr_rep(id_priv->cm_id.ib, &rep);
}

int rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param)
{
	struct rdma_id_private *id_priv;
	int ret;

	id_priv = container_of(id, struct rdma_id_private, id);
	if (!cma_comp(id_priv, CMA_CONNECT))
		return -EINVAL;

	if (!id->qp && conn_param) {
		id_priv->qp_num = conn_param->qp_num;
		id_priv->srq = conn_param->srq;
	}

	switch (rdma_node_get_transport(id->device->node_type)) {
	case RDMA_TRANSPORT_IB:
		if (cma_is_ud_ps(id->ps))
			ret = cma_send_sidr_rep(id_priv, IB_SIDR_SUCCESS,
						conn_param->private_data,
						conn_param->private_data_len);
		else if (conn_param)
			ret = cma_accept_ib(id_priv, conn_param);
		else
			ret = cma_rep_recv(id_priv);
		break;
	case RDMA_TRANSPORT_IWARP:
		ret = cma_accept_iw(id_priv, conn_param);
		break;
	default:
		ret = -ENOSYS;
		break;
	}

	if (ret)
		goto reject;

	return 0;
reject:
	cma_modify_qp_err(id_priv);
	rdma_reject(id, NULL, 0);
	return ret;
}
EXPORT_SYMBOL(rdma_accept);

int rdma_notify(struct rdma_cm_id *id, enum ib_event_type event)
{
	struct rdma_id_private *id_priv;
	int ret;

	id_priv = container_of(id, struct rdma_id_private, id);
	if (!cma_has_cm_dev(id_priv))
		return -EINVAL;

	switch (id->device->node_type) {
	case RDMA_NODE_IB_CA:
		ret = ib_cm_notify(id_priv->cm_id.ib, event);
		break;
	default:
		ret = 0;
		break;
	}
	return ret;
}
EXPORT_SYMBOL(rdma_notify);

int rdma_reject(struct rdma_cm_id *id, const void *private_data,
		u8 private_data_len)
{
	struct rdma_id_private *id_priv;
	int ret;

	id_priv = container_of(id, struct rdma_id_private, id);
	if (!cma_has_cm_dev(id_priv))
		return -EINVAL;

	switch (rdma_node_get_transport(id->device->node_type)) {
	case RDMA_TRANSPORT_IB:
		if (cma_is_ud_ps(id->ps))
			ret = cma_send_sidr_rep(id_priv, IB_SIDR_REJECT,
						private_data, private_data_len);
		else
			ret = ib_send_cm_rej(id_priv->cm_id.ib,
					     IB_CM_REJ_CONSUMER_DEFINED, NULL,
					     0, private_data, private_data_len);
		break;
	case RDMA_TRANSPORT_IWARP:
		ret = iw_cm_reject(id_priv->cm_id.iw,
				   private_data, private_data_len);
		break;
	default:
		ret = -ENOSYS;
		break;
	}
	return ret;
}
EXPORT_SYMBOL(rdma_reject);

int rdma_disconnect(struct rdma_cm_id *id)
{
	struct rdma_id_private *id_priv;
	int ret;

	id_priv = container_of(id, struct rdma_id_private, id);
	if (!cma_has_cm_dev(id_priv))
		return -EINVAL;

	switch (rdma_node_get_transport(id->device->node_type)) {
	case RDMA_TRANSPORT_IB:
		ret = cma_modify_qp_err(id_priv);
		if (ret)
			goto out;
		/* Initiate or respond to a disconnect. */
		if (ib_send_cm_dreq(id_priv->cm_id.ib, NULL, 0))
			ib_send_cm_drep(id_priv->cm_id.ib, NULL, 0);
		break;
	case RDMA_TRANSPORT_IWARP:
		ret = iw_cm_disconnect(id_priv->cm_id.iw, 0);
		break;
	default:
		ret = -EINVAL;
		break;
	}
out:
	return ret;
}
EXPORT_SYMBOL(rdma_disconnect);

static int cma_ib_mc_handler(int status, struct ib_sa_multicast *multicast)
{
	struct rdma_id_private *id_priv;
	struct cma_multicast *mc = multicast->context;
	struct rdma_cm_event event;
	int ret;

	id_priv = mc->id_priv;
	if (cma_disable_remove(id_priv, CMA_ADDR_BOUND) &&
	    cma_disable_remove(id_priv, CMA_ADDR_RESOLVED))
		return 0;

	mutex_lock(&id_priv->qp_mutex);
	if (!status && id_priv->id.qp)
		status = ib_attach_mcast(id_priv->id.qp, &multicast->rec.mgid,
					 multicast->rec.mlid);
	mutex_unlock(&id_priv->qp_mutex);

	memset(&event, 0, sizeof event);
	event.status = status;
	event.param.ud.private_data = mc->context;
	if (!status) {
		event.event = RDMA_CM_EVENT_MULTICAST_JOIN;
		ib_init_ah_from_mcmember(id_priv->id.device,
					 id_priv->id.port_num, &multicast->rec,
					 &event.param.ud.ah_attr);
		event.param.ud.qp_num = 0xFFFFFF;
		event.param.ud.qkey = be32_to_cpu(multicast->rec.qkey);
	} else
		event.event = RDMA_CM_EVENT_MULTICAST_ERROR;

	ret = id_priv->id.event_handler(&id_priv->id, &event);
	if (ret) {
		cma_exch(id_priv, CMA_DESTROYING);
		cma_enable_remove(id_priv);
		rdma_destroy_id(&id_priv->id);
		return 0;
	}

	cma_enable_remove(id_priv);
	return 0;
}

static void cma_set_mgid(struct rdma_id_private *id_priv,
			 struct sockaddr *addr, union ib_gid *mgid)
{
	unsigned char mc_map[MAX_ADDR_LEN];
	struct rdma_dev_addr *dev_addr = &id_priv->id.route.addr.dev_addr;
	struct sockaddr_in *sin = (struct sockaddr_in *) addr;
	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *) addr;

	if (cma_any_addr(addr)) {
		memset(mgid, 0, sizeof *mgid);
	} else if ((addr->sa_family == AF_INET6) &&
		   ((be32_to_cpu(sin6->sin6_addr.s6_addr32[0]) & 0xFF10A01B) ==
								 0xFF10A01B)) {
		/* IPv6 address is an SA assigned MGID. */
		memcpy(mgid, &sin6->sin6_addr, sizeof *mgid);
	} else {
		ip_ib_mc_map(sin->sin_addr.s_addr, mc_map);
		if (id_priv->id.ps == RDMA_PS_UDP)
			mc_map[7] = 0x01;	/* Use RDMA CM signature */
		mc_map[8] = ib_addr_get_pkey(dev_addr) >> 8;
		mc_map[9] = (unsigned char) ib_addr_get_pkey(dev_addr);
		*mgid = *(union ib_gid *) (mc_map + 4);
	}
}

static int cma_join_ib_multicast(struct rdma_id_private *id_priv,
				 struct cma_multicast *mc)
{
	struct ib_sa_mcmember_rec rec;
	struct rdma_dev_addr *dev_addr = &id_pri