Merge: Backport more upstream MANA patches to RHEL 10.2

MR: https://gitlab.com/redhat/centos-stream/src/kernel/centos-stream-10/-/merge_requests/1435

JIRA: https://issues.redhat.com/browse/RHEL-109580

Tested: passed basic tests on an Azure D2s_v6 instance.

Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>

Approved-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Approved-by: John W. Linville <linville@redhat.com>
Approved-by: Michal Schmidt <mschmidt@redhat.com>
Approved-by: CKI KWF Bot <cki-ci-bot+kwf-gitlab-com@redhat.com>

Merged-by: CKI GitLab Kmaint Pipeline Bot <26919896-cki-kmaint-pipeline-bot@users.noreply.gitlab.com>

Author: CKI KWF Bot

drivers/infiniband/hw/mana/counters.c

@@ -32,8 +32,32 @@ static const struct rdma_stat_desc mana_ib_port_stats_desc[] = {
 	[MANA_IB_RATE_INC_EVENTS].name = "rate_inc_events",
 	[MANA_IB_NUM_QPS_RECOVERED].name = "num_qps_recovered",
 	[MANA_IB_CURRENT_RATE].name = "current_rate",
+	[MANA_IB_DUP_RX_REQ].name = "dup_rx_requests",
+	[MANA_IB_TX_BYTES].name = "tx_bytes",
+	[MANA_IB_RX_BYTES].name = "rx_bytes",
+	[MANA_IB_RX_SEND_REQ].name = "rx_send_requests",
+	[MANA_IB_RX_WRITE_REQ].name = "rx_write_requests",
+	[MANA_IB_RX_READ_REQ].name = "rx_read_requests",
+	[MANA_IB_TX_PKT].name = "tx_packets",
+	[MANA_IB_RX_PKT].name = "rx_packets",
 };
 
+static const struct rdma_stat_desc mana_ib_device_stats_desc[] = {
+	[MANA_IB_SENT_CNPS].name = "sent_cnps",
+	[MANA_IB_RECEIVED_ECNS].name = "received_ecns",
+	[MANA_IB_RECEIVED_CNP_COUNT].name = "received_cnp_count",
+	[MANA_IB_QP_CONGESTED_EVENTS].name = "qp_congested_events",
+	[MANA_IB_QP_RECOVERED_EVENTS].name = "qp_recovered_events",
+	[MANA_IB_DEV_RATE_INC_EVENTS].name = "rate_inc_events",
+};
+
+struct rdma_hw_stats *mana_ib_alloc_hw_device_stats(struct ib_device *ibdev)
+{
+	return rdma_alloc_hw_stats_struct(mana_ib_device_stats_desc,
+					  ARRAY_SIZE(mana_ib_device_stats_desc),
+					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
+}
+
 struct rdma_hw_stats *mana_ib_alloc_hw_port_stats(struct ib_device *ibdev,
 						  u32 port_num)
 {
@@ -42,8 +66,39 @@ struct rdma_hw_stats *mana_ib_alloc_hw_port_stats(struct ib_device *ibdev,
 					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
 }
 
-int mana_ib_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
-			 u32 port_num, int index)
+static int mana_ib_get_hw_device_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats)
+{
+	struct mana_ib_dev *mdev = container_of(ibdev, struct mana_ib_dev,
+						ib_dev);
+	struct mana_rnic_query_device_cntrs_resp resp = {};
+	struct mana_rnic_query_device_cntrs_req req = {};
+	int err;
+
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_QUERY_DEVICE_COUNTERS,
+			     sizeof(req), sizeof(resp));
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
+	req.adapter = mdev->adapter_handle;
+
+	err = mana_gd_send_request(mdev_to_gc(mdev), sizeof(req), &req,
+				   sizeof(resp), &resp);
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to query device counters err %d",
+			  err);
+		return err;
+	}
+
+	stats->value[MANA_IB_SENT_CNPS] = resp.sent_cnps;
+	stats->value[MANA_IB_RECEIVED_ECNS] = resp.received_ecns;
+	stats->value[MANA_IB_RECEIVED_CNP_COUNT] = resp.received_cnp_count;
+	stats->value[MANA_IB_QP_CONGESTED_EVENTS] = resp.qp_congested_events;
+	stats->value[MANA_IB_QP_RECOVERED_EVENTS] = resp.qp_recovered_events;
+	stats->value[MANA_IB_DEV_RATE_INC_EVENTS] = resp.rate_inc_events;
+
+	return ARRAY_SIZE(mana_ib_device_stats_desc);
+}
+
+static int mana_ib_get_hw_port_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
+				     u32 port_num)
 {
 	struct mana_ib_dev *mdev = container_of(ibdev, struct mana_ib_dev,
 						ib_dev);
@@ -53,6 +108,7 @@ int mana_ib_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_QUERY_VF_COUNTERS,
 			     sizeof(req), sizeof(resp));
+	req.hdr.resp.msg_version = GDMA_MESSAGE_V2;
 	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 
@@ -101,5 +157,23 @@ int mana_ib_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
 	stats->value[MANA_IB_NUM_QPS_RECOVERED] = resp.num_qps_recovered;
 	stats->value[MANA_IB_CURRENT_RATE] = resp.current_rate;
 
+	stats->value[MANA_IB_DUP_RX_REQ] = resp.dup_rx_req;
+	stats->value[MANA_IB_TX_BYTES] = resp.tx_bytes;
+	stats->value[MANA_IB_RX_BYTES] = resp.rx_bytes;
+	stats->value[MANA_IB_RX_SEND_REQ] = resp.rx_send_req;
+	stats->value[MANA_IB_RX_WRITE_REQ] = resp.rx_write_req;
+	stats->value[MANA_IB_RX_READ_REQ] = resp.rx_read_req;
+	stats->value[MANA_IB_TX_PKT] = resp.tx_pkt;
+	stats->value[MANA_IB_RX_PKT] = resp.rx_pkt;
+
 	return ARRAY_SIZE(mana_ib_port_stats_desc);
 }
+
+int mana_ib_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
+			 u32 port_num, int index)
+{
+	if (!port_num)
+		return mana_ib_get_hw_device_stats(ibdev, stats);
+	else
+		return mana_ib_get_hw_port_stats(ibdev, stats, port_num);
+}

drivers/infiniband/hw/mana/counters.h

@@ -35,10 +35,28 @@ enum mana_ib_port_counters {
 	MANA_IB_RATE_INC_EVENTS,
 	MANA_IB_NUM_QPS_RECOVERED,
 	MANA_IB_CURRENT_RATE,
+	MANA_IB_DUP_RX_REQ,
+	MANA_IB_TX_BYTES,
+	MANA_IB_RX_BYTES,
+	MANA_IB_RX_SEND_REQ,
+	MANA_IB_RX_WRITE_REQ,
+	MANA_IB_RX_READ_REQ,
+	MANA_IB_TX_PKT,
+	MANA_IB_RX_PKT,
+};
+
+enum mana_ib_device_counters {
+	MANA_IB_SENT_CNPS,
+	MANA_IB_RECEIVED_ECNS,
+	MANA_IB_RECEIVED_CNP_COUNT,
+	MANA_IB_QP_CONGESTED_EVENTS,
+	MANA_IB_QP_RECOVERED_EVENTS,
+	MANA_IB_DEV_RATE_INC_EVENTS,
 };
 
 struct rdma_hw_stats *mana_ib_alloc_hw_port_stats(struct ib_device *ibdev,
 						  u32 port_num);
+struct rdma_hw_stats *mana_ib_alloc_hw_device_stats(struct ib_device *ibdev);
 int mana_ib_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
 			 u32 port_num, int index);
 #endif /* _COUNTERS_H_ */

drivers/infiniband/hw/mana/device.c

@@ -65,6 +65,10 @@ static const struct ib_device_ops mana_ib_stats_ops = {
 	.get_hw_stats = mana_ib_get_hw_stats,
 };
 
+static const struct ib_device_ops mana_ib_device_stats_ops = {
+	.alloc_hw_device_stats = mana_ib_alloc_hw_device_stats,
+};
+
 static int mana_ib_netdev_event(struct notifier_block *this,
 				unsigned long event, void *ptr)
 {
@@ -73,28 +77,31 @@ static int mana_ib_netdev_event(struct notifier_block *this,
 	struct gdma_context *gc = dev->gdma_dev->gdma_context;
 	struct mana_context *mc = gc->mana.driver_data;
 	struct net_device *ndev;
+	int i;
 
 	/* Only process events from our parent device */
-	if (event_dev != mc->ports[0])
-		return NOTIFY_DONE;
-
-	switch (event) {
-	case NETDEV_CHANGEUPPER:
-		ndev = mana_get_primary_netdev(mc, 0, &dev->dev_tracker);
-		/*
-		 * RDMA core will setup GID based on updated netdev.
-		 * It's not possible to race with the core as rtnl lock is being
-		 * held.
-		 */
-		ib_device_set_netdev(&dev->ib_dev, ndev, 1);
-
-		/* mana_get_primary_netdev() returns ndev with refcount held */
-		netdev_put(ndev, &dev->dev_tracker);
-
-		return NOTIFY_OK;
-	default:
-		return NOTIFY_DONE;
-	}
+	for (i = 0; i < dev->ib_dev.phys_port_cnt; i++)
+		if (event_dev == mc->ports[i]) {
+			switch (event) {
+			case NETDEV_CHANGEUPPER:
+				ndev = mana_get_primary_netdev(mc, i, &dev->dev_tracker);
+				/*
+				 * RDMA core will setup GID based on updated netdev.
+				 * It's not possible to race with the core as rtnl lock is being
+				 * held.
+				 */
+				ib_device_set_netdev(&dev->ib_dev, ndev, i + 1);
+
+				/* mana_get_primary_netdev() returns ndev with refcount held */
+				if (ndev)
+					netdev_put(ndev, &dev->dev_tracker);
+
+				return NOTIFY_OK;
+			default:
+				return NOTIFY_DONE;
+			}
+		}
+	return NOTIFY_DONE;
 }
 
 static int mana_ib_probe(struct auxiliary_device *adev,
@@ -107,7 +114,7 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 	struct net_device *ndev;
 	struct mana_ib_dev *dev;
 	u8 mac_addr[ETH_ALEN];
-	int ret;
+	int ret, i;
 
 	dev = ib_alloc_device(mana_ib_dev, ib_dev);
 	if (!dev)
@@ -122,51 +129,56 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 
 	if (mana_ib_is_rnic(dev)) {
 		dev->ib_dev.phys_port_cnt = 1;
-		ndev = mana_get_primary_netdev(mc, 0, &dev->dev_tracker);
-		if (!ndev) {
-			ret = -ENODEV;
-			ibdev_err(&dev->ib_dev, "Failed to get netdev for IB port 1");
-			goto free_ib_device;
-		}
-		ether_addr_copy(mac_addr, ndev->dev_addr);
-		addrconf_addr_eui48((u8 *)&dev->ib_dev.node_guid, ndev->dev_addr);
-		ret = ib_device_set_netdev(&dev->ib_dev, ndev, 1);
-		/* mana_get_primary_netdev() returns ndev with refcount held */
-		netdev_put(ndev, &dev->dev_tracker);
-		if (ret) {
-			ibdev_err(&dev->ib_dev, "Failed to set ib netdev, ret %d", ret);
-			goto free_ib_device;
-		}
-
-		dev->nb.notifier_call = mana_ib_netdev_event;
-		ret = register_netdevice_notifier(&dev->nb);
-		if (ret) {
-			ibdev_err(&dev->ib_dev, "Failed to register net notifier, %d",
-				  ret);
-			goto free_ib_device;
-		}
-
+		addrconf_addr_eui48((u8 *)&dev->ib_dev.node_guid, mc->ports[0]->dev_addr);
 		ret = mana_ib_gd_query_adapter_caps(dev);
 		if (ret) {
 			ibdev_err(&dev->ib_dev, "Failed to query device caps, ret %d", ret);
-			goto deregister_net_notifier;
+			goto free_ib_device;
 		}
 
 		ib_set_device_ops(&dev->ib_dev, &mana_ib_stats_ops);
+		if (dev->adapter_caps.feature_flags & MANA_IB_FEATURE_DEV_COUNTERS_SUPPORT)
+			ib_set_device_ops(&dev->ib_dev, &mana_ib_device_stats_ops);
 
 		ret = mana_ib_create_eqs(dev);
 		if (ret) {
 			ibdev_err(&dev->ib_dev, "Failed to create EQs, ret %d", ret);
-			goto deregister_net_notifier;
+			goto free_ib_device;
 		}
 
 		ret = mana_ib_gd_create_rnic_adapter(dev);
 		if (ret)
 			goto destroy_eqs;
 
-		ret = mana_ib_gd_config_mac(dev, ADDR_OP_ADD, mac_addr);
+		if (dev->adapter_caps.feature_flags & MANA_IB_FEATURE_MULTI_PORTS_SUPPORT)
+			dev->ib_dev.phys_port_cnt = mc->num_ports;
+
+		for (i = 0; i < dev->ib_dev.phys_port_cnt; i++) {
+			ndev = mana_get_primary_netdev(mc, i, &dev->dev_tracker);
+			if (!ndev) {
+				ret = -ENODEV;
+				ibdev_err(&dev->ib_dev,
+					  "Failed to get netdev for IB port %d", i + 1);
+				goto destroy_rnic;
+			}
+			ether_addr_copy(mac_addr, ndev->dev_addr);
+			ret = ib_device_set_netdev(&dev->ib_dev, ndev, i + 1);
+			/* mana_get_primary_netdev() returns ndev with refcount held */
+			netdev_put(ndev, &dev->dev_tracker);
+			if (ret) {
+				ibdev_err(&dev->ib_dev, "Failed to set ib netdev, ret %d", ret);
+				goto destroy_rnic;
+			}
+			ret = mana_ib_gd_config_mac(dev, ADDR_OP_ADD, mac_addr);
+			if (ret) {
+				ibdev_err(&dev->ib_dev, "Failed to add Mac address, ret %d", ret);
+				goto destroy_rnic;
+			}
+		}
+		dev->nb.notifier_call = mana_ib_netdev_event;
+		ret = register_netdevice_notifier(&dev->nb);
 		if (ret) {
-			ibdev_err(&dev->ib_dev, "Failed to add Mac address, ret %d", ret);
+			ibdev_err(&dev->ib_dev, "Failed to register net notifier, %d", ret);
 			goto destroy_rnic;
 		}
 	} else {
@@ -182,7 +194,7 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 				       MANA_AV_BUFFER_SIZE, 0);
 	if (!dev->av_pool) {
 		ret = -ENOMEM;
-		goto destroy_rnic;
+		goto deregister_net_notifier;
 	}
 
 	ibdev_dbg(&dev->ib_dev, "mdev=%p id=%d num_ports=%d\n", mdev,
@@ -199,15 +211,15 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 
 deallocate_pool:
 	dma_pool_destroy(dev->av_pool);
+deregister_net_notifier:
+	if (mana_ib_is_rnic(dev))
+		unregister_netdevice_notifier(&dev->nb);
 destroy_rnic:
 	if (mana_ib_is_rnic(dev))
 		mana_ib_gd_destroy_rnic_adapter(dev);
 destroy_eqs:
 	if (mana_ib_is_rnic(dev))
 		mana_ib_destroy_eqs(dev);
-deregister_net_notifier:
-	if (mana_ib_is_rnic(dev))
-		unregister_netdevice_notifier(&dev->nb);
 free_ib_device:
 	xa_destroy(&dev->qp_table_wq);
 	ib_dealloc_device(&dev->ib_dev);
@@ -221,9 +233,9 @@ static void mana_ib_remove(struct auxiliary_device *adev)
 	ib_unregister_device(&dev->ib_dev);
 	dma_pool_destroy(dev->av_pool);
 	if (mana_ib_is_rnic(dev)) {
+		unregister_netdevice_notifier(&dev->nb);
 		mana_ib_gd_destroy_rnic_adapter(dev);
 		mana_ib_destroy_eqs(dev);
-		unregister_netdevice_notifier(&dev->nb);
 	}
 	xa_destroy(&dev->qp_table_wq);
 	ib_dealloc_device(&dev->ib_dev);

drivers/infiniband/hw/mana/main.c

@@ -563,8 +563,14 @@ int mana_ib_get_port_immutable(struct ib_device *ibdev, u32 port_num,
 	immutable->gid_tbl_len = attr.gid_tbl_len;
 
 	if (mana_ib_is_rnic(dev)) {
-		immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
-		immutable->max_mad_size = IB_MGMT_MAD_SIZE;
+		if (port_num == 1) {
+			immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
+			immutable->max_mad_size = IB_MGMT_MAD_SIZE;
+		} else {
+			immutable->core_cap_flags = RDMA_CORE_CAP_PROT_ROCE_UDP_ENCAP
+						    | RDMA_CORE_CAP_ETH_AH;
+			immutable->max_mad_size = 0;
+		}
 	} else {
 		immutable->core_cap_flags = RDMA_CORE_PORT_RAW_PACKET;
 	}
@@ -633,8 +639,9 @@ int mana_ib_query_port(struct ib_device *ibdev, u32 port,
 	props->pkey_tbl_len = 1;
 	if (mana_ib_is_rnic(dev)) {
 		props->gid_tbl_len = 16;
-		props->port_cap_flags = IB_PORT_CM_SUP;
 		props->ip_gids = true;
+		if (port == 1)
+			props->port_cap_flags = IB_PORT_CM_SUP;
 	}
 
 	return 0;