mlx4_core: Multiple port type support

Multi-protocol adapters support different port types.  Each consumer
of mlx4_core queries for supported port types; in particular mlx4_ib
can no longer assume that all physical ports belong to it.  Port type
is configured through a sysfs interface.  When the type of a port is
changed, all mlx4 interfaces are unregistered, and then registered
again with the new port types.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

drivers/infiniband/hw/mlx4/mad.c

@@ -298,7 +298,7 @@ int mlx4_ib_mad_init(struct mlx4_ib_dev *dev)
 	int p, q;
 	int ret;
 
-	for (p = 0; p < dev->dev->caps.num_ports; ++p)
+	for (p = 0; p < dev->num_ports; ++p)
 		for (q = 0; q <= 1; ++q) {
 			agent = ib_register_mad_agent(&dev->ib_dev, p + 1,
 						      q ? IB_QPT_GSI : IB_QPT_SMI,
@@ -314,7 +314,7 @@ int mlx4_ib_mad_init(struct mlx4_ib_dev *dev)
 	return 0;
 
 err:
-	for (p = 0; p < dev->dev->caps.num_ports; ++p)
+	for (p = 0; p < dev->num_ports; ++p)
 		for (q = 0; q <= 1; ++q)
 			if (dev->send_agent[p][q])
 				ib_unregister_mad_agent(dev->send_agent[p][q]);
@@ -327,7 +327,7 @@ void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev)
 	struct ib_mad_agent *agent;
 	int p, q;
 
-	for (p = 0; p < dev->dev->caps.num_ports; ++p) {
+	for (p = 0; p < dev->num_ports; ++p) {
 		for (q = 0; q <= 1; ++q) {
 			agent = dev->send_agent[p][q];
 			dev->send_agent[p][q] = NULL;

drivers/infiniband/hw/mlx4/main.c

@@ -574,7 +574,10 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	ibdev->ib_dev.owner		= THIS_MODULE;
 	ibdev->ib_dev.node_type		= RDMA_NODE_IB_CA;
 	ibdev->ib_dev.local_dma_lkey	= dev->caps.reserved_lkey;
-	ibdev->ib_dev.phys_port_cnt	= dev->caps.num_ports;
+	ibdev->num_ports = 0;
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB)
+		ibdev->num_ports++;
+	ibdev->ib_dev.phys_port_cnt     = ibdev->num_ports;
 	ibdev->ib_dev.num_comp_vectors	= 1;
 	ibdev->ib_dev.dma_device	= &dev->pdev->dev;
 
@@ -691,7 +694,7 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
 	struct mlx4_ib_dev *ibdev = ibdev_ptr;
 	int p;
 
-	for (p = 1; p <= dev->caps.num_ports; ++p)
+	for (p = 1; p <= ibdev->num_ports; ++p)
 		mlx4_CLOSE_PORT(dev, p);
 
 	mlx4_ib_mad_cleanup(ibdev);
@@ -706,6 +709,10 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
 			  enum mlx4_dev_event event, int port)
 {
 	struct ib_event ibev;
+	struct mlx4_ib_dev *ibdev = to_mdev((struct ib_device *) ibdev_ptr);
+
+	if (port > ibdev->num_ports)
+		return;
 
 	switch (event) {
 	case MLX4_DEV_EVENT_PORT_UP:

drivers/infiniband/hw/mlx4/mlx4_ib.h

@@ -162,6 +162,7 @@ struct mlx4_ib_ah {
 struct mlx4_ib_dev {
 	struct ib_device	ib_dev;
 	struct mlx4_dev	       *dev;
+	int			num_ports;
 	void __iomem	       *uar_map;
 
 	struct mlx4_uar		priv_uar;

drivers/net/mlx4/fw.c

@@ -88,6 +88,7 @@ static void dump_dev_cap_flags(struct mlx4_dev *dev, u32 flags)
 		[ 8] = "P_Key violation counter",
 		[ 9] = "Q_Key violation counter",
 		[10] = "VMM",
+		[12] = "DPDP",
 		[16] = "MW support",
 		[17] = "APM support",
 		[18] = "Atomic ops support",
@@ -354,6 +355,7 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 			dev_cap->max_pkeys[i]	   = 1 << (field & 0xf);
 		}
 	} else {
+#define QUERY_PORT_SUPPORTED_TYPE_OFFSET	0x00
 #define QUERY_PORT_MTU_OFFSET			0x01
 #define QUERY_PORT_ETH_MTU_OFFSET		0x02
 #define QUERY_PORT_WIDTH_OFFSET			0x06
@@ -368,6 +370,8 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 			if (err)
 				goto out;
 
+			MLX4_GET(field, outbox, QUERY_PORT_SUPPORTED_TYPE_OFFSET);
+			dev_cap->supported_port_types[i] = field & 3;
 			MLX4_GET(field, outbox, QUERY_PORT_MTU_OFFSET);
 			dev_cap->ib_mtu[i]	   = field & 0xf;
 			MLX4_GET(field, outbox, QUERY_PORT_WIDTH_OFFSET);

drivers/net/mlx4/fw.h

@@ -104,6 +104,7 @@ struct mlx4_dev_cap {
 	u32 reserved_lkey;
 	u64 max_icm_sz;
 	int max_gso_sz;
+	u8  supported_port_types[MLX4_MAX_PORTS + 1];
 	u8  log_max_macs[MLX4_MAX_PORTS + 1];
 	u8  log_max_vlans[MLX4_MAX_PORTS + 1];
 };

drivers/net/mlx4/main.c

@@ -98,6 +98,44 @@ module_param_named(use_prio, use_prio, bool, 0444);
 MODULE_PARM_DESC(use_prio, "Enable steering by VLAN priority on ETH ports "
 		  "(0/1, default 0)");
 
+static int mlx4_check_port_params(struct mlx4_dev *dev,
+				  enum mlx4_port_type *port_type)
+{
+	int i;
+
+	for (i = 0; i < dev->caps.num_ports - 1; i++) {
+		if (port_type[i] != port_type[i+1] &&
+		    !(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
+			mlx4_err(dev, "Only same port types supported "
+				 "on this HCA, aborting.\n");
+			return -EINVAL;
+		}
+	}
+	if ((port_type[0] == MLX4_PORT_TYPE_ETH) &&
+	    (port_type[1] == MLX4_PORT_TYPE_IB)) {
+		mlx4_err(dev, "eth-ib configuration is not supported.\n");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < dev->caps.num_ports; i++) {
+		if (!(port_type[i] & dev->caps.supported_type[i+1])) {
+			mlx4_err(dev, "Requested port type for port %d is not "
+				      "supported on this HCA\n", i + 1);
+			return -EINVAL;
+		}
+	}
+	return 0;
+}
+
+static void mlx4_set_port_mask(struct mlx4_dev *dev)
+{
+	int i;
+
+	dev->caps.port_mask = 0;
+	for (i = 1; i <= dev->caps.num_ports; ++i)
+		if (dev->caps.port_type[i] == MLX4_PORT_TYPE_IB)
+			dev->caps.port_mask |= 1 << (i - 1);
+}
 static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 {
 	int err;
@@ -139,6 +177,7 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 		dev->caps.port_width_cap[i] = dev_cap->max_port_width[i];
 		dev->caps.eth_mtu_cap[i]    = dev_cap->eth_mtu[i];
 		dev->caps.def_mac[i]        = dev_cap->def_mac[i];
+		dev->caps.supported_type[i] = dev_cap->supported_port_types[i];
 	}
 
 	dev->caps.num_uars	     = dev_cap->uar_size / PAGE_SIZE;
@@ -182,6 +221,11 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 	dev->caps.log_num_prios = use_prio ? 3 : 0;
 
 	for (i = 1; i <= dev->caps.num_ports; ++i) {
+		if (dev->caps.supported_type[i] != MLX4_PORT_TYPE_ETH)
+			dev->caps.port_type[i] = MLX4_PORT_TYPE_IB;
+		else
+			dev->caps.port_type[i] = MLX4_PORT_TYPE_ETH;
+
 		if (dev->caps.log_num_macs > dev_cap->log_max_macs[i]) {
 			dev->caps.log_num_macs = dev_cap->log_max_macs[i];
 			mlx4_warn(dev, "Requested number of MACs is too much "
@@ -196,6 +240,8 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 		}
 	}
 
+	mlx4_set_port_mask(dev);
+
 	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW] = dev_cap->reserved_qps;
 	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_ETH_ADDR] =
 		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_ADDR] =
@@ -213,6 +259,95 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 	return 0;
 }
 
+/*
+ * Change the port configuration of the device.
+ * Every user of this function must hold the port mutex.
+ */
+static int mlx4_change_port_types(struct mlx4_dev *dev,
+				  enum mlx4_port_type *port_types)
+{
+	int err = 0;
+	int change = 0;
+	int port;
+
+	for (port = 0; port <  dev->caps.num_ports; port++) {
+		if (port_types[port] != dev->caps.port_type[port + 1]) {
+			change = 1;
+			dev->caps.port_type[port + 1] = port_types[port];
+		}
+	}
+	if (change) {
+		mlx4_unregister_device(dev);
+		for (port = 1; port <= dev->caps.num_ports; port++) {
+			mlx4_CLOSE_PORT(dev, port);
+			err = mlx4_SET_PORT(dev, port);
+			if (err) {
+				mlx4_err(dev, "Failed to set port %d, "
+					      "aborting\n", port);
+				goto out;
+			}
+		}
+		mlx4_set_port_mask(dev);
+		err = mlx4_register_device(dev);
+	}
+
+out:
+	return err;
+}
+
+static ssize_t show_port_type(struct device *dev,
+			      struct device_attribute *attr,
+			      char *buf)
+{
+	struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
+						   port_attr);
+	struct mlx4_dev *mdev = info->dev;
+
+	return sprintf(buf, "%s\n",
+		       mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_IB ?
+		       "ib" : "eth");
+}
+
+static ssize_t set_port_type(struct device *dev,
+			     struct device_attribute *attr,
+			     const char *buf, size_t count)
+{
+	struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
+						   port_attr);
+	struct mlx4_dev *mdev = info->dev;
+	struct mlx4_priv *priv = mlx4_priv(mdev);
+	enum mlx4_port_type types[MLX4_MAX_PORTS];
+	int i;
+	int err = 0;
+
+	if (!strcmp(buf, "ib\n"))
+		info->tmp_type = MLX4_PORT_TYPE_IB;
+	else if (!strcmp(buf, "eth\n"))
+		info->tmp_type = MLX4_PORT_TYPE_ETH;
+	else {
+		mlx4_err(mdev, "%s is not supported port type\n", buf);
+		return -EINVAL;
+	}
+
+	mutex_lock(&priv->port_mutex);
+	for (i = 0; i < mdev->caps.num_ports; i++)
+		types[i] = priv->port[i+1].tmp_type ? priv->port[i+1].tmp_type :
+					mdev->caps.port_type[i+1];
+
+	err = mlx4_check_port_params(mdev, types);
+	if (err)
+		goto out;
+
+	for (i = 1; i <= mdev->caps.num_ports; i++)
+		priv->port[i].tmp_type = 0;
+
+	err = mlx4_change_port_types(mdev, types);
+
+out:
+	mutex_unlock(&priv->port_mutex);
+	return err ? err : count;
+}
+
 static int mlx4_load_fw(struct mlx4_dev *dev)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
@@ -617,6 +752,7 @@ static int mlx4_setup_hca(struct mlx4_dev *dev)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
 	int err;
+	int port;
 
 	err = mlx4_init_uar_table(dev);
 	if (err) {
@@ -715,8 +851,20 @@ static int mlx4_setup_hca(struct mlx4_dev *dev)
 		goto err_qp_table_free;
 	}
 
+	for (port = 1; port <= dev->caps.num_ports; port++) {
+		err = mlx4_SET_PORT(dev, port);
+		if (err) {
+			mlx4_err(dev, "Failed to set port %d, aborting\n",
+				port);
+			goto err_mcg_table_free;
+		}
+	}
+
 	return 0;
 
+err_mcg_table_free:
+	mlx4_cleanup_mcg_table(dev);
+
 err_qp_table_free:
 	mlx4_cleanup_qp_table(dev);
 
@@ -780,14 +928,37 @@ no_msi:
 		priv->eq_table.eq[i].irq = dev->pdev->irq;
 }
 
-static void mlx4_init_port_info(struct mlx4_dev *dev, int port)
+static int mlx4_init_port_info(struct mlx4_dev *dev, int port)
 {
 	struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
+	int err = 0;
 
 	info->dev = dev;
 	info->port = port;
 	mlx4_init_mac_table(dev, &info->mac_table);
 	mlx4_init_vlan_table(dev, &info->vlan_table);
+
+	sprintf(info->dev_name, "mlx4_port%d", port);
+	info->port_attr.attr.name = info->dev_name;
+	info->port_attr.attr.mode = S_IRUGO | S_IWUSR;
+	info->port_attr.show      = show_port_type;
+	info->port_attr.store     = set_port_type;
+
+	err = device_create_file(&dev->pdev->dev, &info->port_attr);
+	if (err) {
+		mlx4_err(dev, "Failed to create file for port %d\n", port);
+		info->port = -1;
+	}
+
+	return err;
+}
+
+static void mlx4_cleanup_port_info(struct mlx4_port_info *info)
+{
+	if (info->port < 0)
+		return;
+
+	device_remove_file(&info->dev->pdev->dev, &info->port_attr);
 }
 
 static int __mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
@@ -870,6 +1041,8 @@ static int __mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	INIT_LIST_HEAD(&priv->ctx_list);
 	spin_lock_init(&priv->ctx_lock);
 
+	mutex_init(&priv->port_mutex);
+
 	INIT_LIST_HEAD(&priv->pgdir_list);
 	mutex_init(&priv->pgdir_mutex);
 
@@ -905,18 +1078,24 @@ static int __mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	if (err)
 		goto err_close;
 
-	for (port = 1; port <= dev->caps.num_ports; port++)
+	for (port = 1; port <= dev->caps.num_ports; port++) {
-		mlx4_init_port_info(dev, port);
+		err = mlx4_init_port_info(dev, port);
+		if (err)
+			goto err_port;
+	}
 
 	err = mlx4_register_device(dev);
 	if (err)
-		goto err_cleanup;
+		goto err_port;
 
 	pci_set_drvdata(pdev, dev);
 
 	return 0;
 
-err_cleanup:
+err_port:
+	for (port = 1; port <= dev->caps.num_ports; port++)
+		mlx4_cleanup_port_info(&priv->port[port]);
+
 	mlx4_cleanup_mcg_table(dev);
 	mlx4_cleanup_qp_table(dev);
 	mlx4_cleanup_srq_table(dev);
@@ -973,8 +1152,10 @@ static void mlx4_remove_one(struct pci_dev *pdev)
 	if (dev) {
 		mlx4_unregister_device(dev);
 
-		for (p = 1; p <= dev->caps.num_ports; ++p)
+		for (p = 1; p <= dev->caps.num_ports; p++) {
+			mlx4_cleanup_port_info(&priv->port[p]);
 			mlx4_CLOSE_PORT(dev, p);
+		}
 
 		mlx4_cleanup_mcg_table(dev);
 		mlx4_cleanup_qp_table(dev);
@@ -1026,10 +1207,28 @@ static struct pci_driver mlx4_driver = {
 	.remove		= __devexit_p(mlx4_remove_one)
 };
 
+static int __init mlx4_verify_params(void)
+{
+	if ((log_num_mac < 0) || (log_num_mac > 7)) {
+		printk(KERN_WARNING "mlx4_core: bad num_mac: %d\n", log_num_mac);
+		return -1;
+	}
+
+	if ((log_num_vlan < 0) || (log_num_vlan > 7)) {
+		printk(KERN_WARNING "mlx4_core: bad num_vlan: %d\n", log_num_vlan);
+		return -1;
+	}
+
+	return 0;
+}
+
 static int __init mlx4_init(void)
 {
 	int ret;
 
+	if (mlx4_verify_params())
+		return -EINVAL;
+
 	ret = mlx4_catas_init();
 	if (ret)
 		return ret;

drivers/net/mlx4/mlx4.h

@@ -277,6 +277,9 @@ struct mlx4_vlan_table {
 struct mlx4_port_info {
 	struct mlx4_dev	       *dev;
 	int			port;
+	char			dev_name[16];
+	struct device_attribute port_attr;
+	enum mlx4_port_type	tmp_type;
 	struct mlx4_mac_table	mac_table;
 	struct mlx4_vlan_table	vlan_table;
 };
@@ -310,6 +313,7 @@ struct mlx4_priv {
 	struct mlx4_uar		driver_uar;
 	void __iomem	       *kar;
 	struct mlx4_port_info	port[MLX4_MAX_PORTS + 1];
+	struct mutex		port_mutex;
 };
 
 static inline struct mlx4_priv *mlx4_priv(struct mlx4_dev *dev)
@@ -383,4 +387,6 @@ void mlx4_handle_catas_err(struct mlx4_dev *dev);
 void mlx4_init_mac_table(struct mlx4_dev *dev, struct mlx4_mac_table *table);
 void mlx4_init_vlan_table(struct mlx4_dev *dev, struct mlx4_vlan_table *table);
 
+int mlx4_SET_PORT(struct mlx4_dev *dev, u8 port);
+
 #endif /* MLX4_H */

drivers/net/mlx4/port.c

@@ -257,3 +257,26 @@ out:
 	mutex_unlock(&table->mutex);
 }
 EXPORT_SYMBOL_GPL(mlx4_unregister_vlan);
+
+int mlx4_SET_PORT(struct mlx4_dev *dev, u8 port)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+	u8 is_eth = dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	memset(mailbox->buf, 0, 256);
+	if (is_eth) {
+		((u8 *) mailbox->buf)[3] = 6;
+		((__be16 *) mailbox->buf)[4] = cpu_to_be16(1 << 15);
+		((__be16 *) mailbox->buf)[6] = cpu_to_be16(1 << 15);
+	}
+	err = mlx4_cmd(dev, mailbox->dma, port, is_eth, MLX4_CMD_SET_PORT,
+		       MLX4_CMD_TIME_CLASS_B);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}

include/linux/mlx4/device.h

@@ -60,6 +60,7 @@ enum {
 	MLX4_DEV_CAP_FLAG_IPOIB_CSUM	= 1 <<  7,
 	MLX4_DEV_CAP_FLAG_BAD_PKEY_CNTR	= 1 <<  8,
 	MLX4_DEV_CAP_FLAG_BAD_QKEY_CNTR	= 1 <<  9,
+	MLX4_DEV_CAP_FLAG_DPDP		= 1 << 12,
 	MLX4_DEV_CAP_FLAG_MEM_WINDOW	= 1 << 16,
 	MLX4_DEV_CAP_FLAG_APM		= 1 << 17,
 	MLX4_DEV_CAP_FLAG_ATOMIC	= 1 << 18,
@@ -153,6 +154,11 @@ enum mlx4_qp_region {
 	MLX4_NUM_QP_REGION
 };
 
+enum mlx4_port_type {
+	MLX4_PORT_TYPE_IB	= 1 << 0,
+	MLX4_PORT_TYPE_ETH	= 1 << 1,
+};
+
 enum mlx4_special_vlan_idx {
 	MLX4_NO_VLAN_IDX        = 0,
 	MLX4_VLAN_MISS_IDX,
@@ -226,6 +232,9 @@ struct mlx4_caps {
 	int                     log_num_macs;
 	int                     log_num_vlans;
 	int                     log_num_prios;
+	enum mlx4_port_type	port_type[MLX4_MAX_PORTS + 1];
+	u8			supported_type[MLX4_MAX_PORTS + 1];
+	u32			port_mask;
 };
 
 struct mlx4_buf_list {
@@ -380,6 +389,11 @@ struct mlx4_init_port_param {
 	u64			si_guid;
 };
 
+#define mlx4_foreach_port(port, dev, type)				\
+	for ((port) = 1; (port) <= (dev)->caps.num_ports; (port)++)	\
+		if (((type) == MLX4_PORT_TYPE_IB ? (dev)->caps.port_mask : \
+		     ~(dev)->caps.port_mask) & 1 << ((port) - 1))
+
 int mlx4_buf_alloc(struct mlx4_dev *dev, int size, int max_direct,
 		   struct mlx4_buf *buf);
 void mlx4_buf_free(struct mlx4_dev *dev, int size, struct mlx4_buf *buf);