Merge tag 'mlx5-updates-2019-11-12' of git://git.kernel.org/pub/scm/l…

…inux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5-updates-2019-11-12

1) Merge mlx5-next for devlink reload and flowtable offloads dependencies
2) Devlink reload support
3) TC Flowtable offloads
4) Misc cleanup
====================

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

Documentation/networking/device_drivers/mellanox/mlx5.rst

@@ -154,6 +154,27 @@ User command examples:
       values:
          cmode runtime value smfs
 
+enable_roce: RoCE enablement state
+----------------------------------
+RoCE enablement state controls driver support for RoCE traffic.
+When RoCE is disabled, there is no gid table, only raw ethernet QPs are supported and traffic on the well known UDP RoCE port is handled as raw ethernet traffic.
+
+To change RoCE enablement state a user must change the driverinit cmode value and run devlink reload.
+
+User command examples:
+
+- Disable RoCE::
+
+    $ devlink dev param set pci/0000:06:00.0 name enable_roce value false cmode driverinit
+    $ devlink dev reload pci/0000:06:00.0
+
+- Read RoCE enablement state::
+
+    $ devlink dev param show pci/0000:06:00.0 name enable_roce
+      pci/0000:06:00.0:
+      name enable_roce type generic
+      values:
+         cmode driverinit value true
 
 Devlink health reporters
 ========================

Documentation/networking/devlink-params-mlx5.txt

@@ -0,0 +1,17 @@
+flow_steering_mode	[DEVICE, DRIVER-SPECIFIC]
+			Controls the flow steering mode of the driver.
+			Two modes are supported:
+			1. 'dmfs' - Device managed flow steering.
+			2. 'smfs  - Software/Driver managed flow steering.
+			In DMFS mode, the HW steering entities are created and
+			managed through the Firmware.
+			In SMFS mode, the HW steering entities are created and
+			managed though by the driver directly into Hardware
+			without firmware intervention.
+			Type: String
+			Configuration mode: runtime
+
+enable_roce		[DEVICE, GENERIC]
+			Enable handling of RoCE traffic in the device.
+			Defaultly enabled.
+			Configuration mode: driverinit

Documentation/networking/devlink-params.txt

@@ -65,3 +65,7 @@ reset_dev_on_drv_probe	[DEVICE, GENERIC]
 			  Reset only if device firmware can be found in the
 			  filesystem.
 			Type: u8
+
+enable_roce		[DEVICE, GENERIC]
+			Enable handling of RoCE traffic in the device.
+			Type: Boolean

drivers/infiniband/hw/mlx5/ib_rep.c

@@ -35,7 +35,7 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 	int vport_index;
 
 	if (rep->vport == MLX5_VPORT_UPLINK)
-		profile = &uplink_rep_profile;
+		profile = &raw_eth_profile;
 	else
 		return mlx5_ib_set_vport_rep(dev, rep);
 

drivers/infiniband/hw/mlx5/ib_rep.h

@@ -10,7 +10,7 @@
 #include "mlx5_ib.h"
 
 #ifdef CONFIG_MLX5_ESWITCH
-extern const struct mlx5_ib_profile uplink_rep_profile;
+extern const struct mlx5_ib_profile raw_eth_profile;
 
 u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw);
 struct mlx5_ib_dev *mlx5_ib_get_rep_ibdev(struct mlx5_eswitch *esw,

drivers/infiniband/hw/mlx5/main.c

@@ -1031,7 +1031,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 	if (MLX5_CAP_GEN(mdev, cd))
 		props->device_cap_flags |= IB_DEVICE_CROSS_CHANNEL;
 
-	if (!mlx5_core_is_pf(mdev))
+	if (mlx5_core_is_vf(mdev))
 		props->device_cap_flags |= IB_DEVICE_VIRTUAL_FUNCTION;
 
 	if (mlx5_ib_port_link_layer(ibdev, 1) ==
@@ -5145,8 +5145,7 @@ static int mlx5_port_immutable(struct ib_device *ibdev, u8 port_num,
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
 	immutable->core_cap_flags = get_core_cap_flags(ibdev, &rep);
-	if ((ll == IB_LINK_LAYER_INFINIBAND) || MLX5_CAP_GEN(dev->mdev, roce))
-		immutable->max_mad_size = IB_MGMT_MAD_SIZE;
+	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 
 	return 0;
 }
@@ -5249,11 +5248,9 @@ static int mlx5_enable_eth(struct mlx5_ib_dev *dev)
 {
 	int err;
 
-	if (MLX5_CAP_GEN(dev->mdev, roce)) {
-		err = mlx5_nic_vport_enable_roce(dev->mdev);
-		if (err)
-			return err;
-	}
+	err = mlx5_nic_vport_enable_roce(dev->mdev);
+	if (err)
+		return err;
 
 	err = mlx5_eth_lag_init(dev);
 	if (err)
@@ -5262,17 +5259,15 @@ static int mlx5_enable_eth(struct mlx5_ib_dev *dev)
 	return 0;
 
 err_disable_roce:
-	if (MLX5_CAP_GEN(dev->mdev, roce))
-		mlx5_nic_vport_disable_roce(dev->mdev);
+	mlx5_nic_vport_disable_roce(dev->mdev);
 
 	return err;
 }
 
 static void mlx5_disable_eth(struct mlx5_ib_dev *dev)
 {
 	mlx5_eth_lag_cleanup(dev);
-	if (MLX5_CAP_GEN(dev->mdev, roce))
-		mlx5_nic_vport_disable_roce(dev->mdev);
+	mlx5_nic_vport_disable_roce(dev->mdev);
 }
 
 struct mlx5_ib_counter {
@@ -6444,7 +6439,7 @@ static const struct ib_device_ops mlx5_ib_dev_port_rep_ops = {
 	.query_port = mlx5_ib_rep_query_port,
 };
 
-static int mlx5_ib_stage_rep_non_default_cb(struct mlx5_ib_dev *dev)
+static int mlx5_ib_stage_raw_eth_non_default_cb(struct mlx5_ib_dev *dev)
 {
 	ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_port_rep_ops);
 	return 0;
@@ -6484,7 +6479,7 @@ static void mlx5_ib_stage_common_roce_cleanup(struct mlx5_ib_dev *dev)
 	mlx5_remove_netdev_notifier(dev, port_num);
 }
 
-static int mlx5_ib_stage_rep_roce_init(struct mlx5_ib_dev *dev)
+static int mlx5_ib_stage_raw_eth_roce_init(struct mlx5_ib_dev *dev)
 {
 	struct mlx5_core_dev *mdev = dev->mdev;
 	enum rdma_link_layer ll;
@@ -6500,7 +6495,7 @@ static int mlx5_ib_stage_rep_roce_init(struct mlx5_ib_dev *dev)
 	return err;
 }
 
-static void mlx5_ib_stage_rep_roce_cleanup(struct mlx5_ib_dev *dev)
+static void mlx5_ib_stage_raw_eth_roce_cleanup(struct mlx5_ib_dev *dev)
 {
 	mlx5_ib_stage_common_roce_cleanup(dev);
 }
@@ -6807,7 +6802,7 @@ static const struct mlx5_ib_profile pf_profile = {
 		     mlx5_ib_stage_delay_drop_cleanup),
 };
 
-const struct mlx5_ib_profile uplink_rep_profile = {
+const struct mlx5_ib_profile raw_eth_profile = {
 	STAGE_CREATE(MLX5_IB_STAGE_INIT,
 		     mlx5_ib_stage_init_init,
 		     mlx5_ib_stage_init_cleanup),
@@ -6818,11 +6813,11 @@ const struct mlx5_ib_profile uplink_rep_profile = {
 		     mlx5_ib_stage_caps_init,
 		     NULL),
 	STAGE_CREATE(MLX5_IB_STAGE_NON_DEFAULT_CB,
-		     mlx5_ib_stage_rep_non_default_cb,
+		     mlx5_ib_stage_raw_eth_non_default_cb,
 		     NULL),
 	STAGE_CREATE(MLX5_IB_STAGE_ROCE,
-		     mlx5_ib_stage_rep_roce_init,
-		     mlx5_ib_stage_rep_roce_cleanup),
+		     mlx5_ib_stage_raw_eth_roce_init,
+		     mlx5_ib_stage_raw_eth_roce_cleanup),
 	STAGE_CREATE(MLX5_IB_STAGE_SRQ,
 		     mlx5_init_srq_table,
 		     mlx5_cleanup_srq_table),
@@ -6898,6 +6893,7 @@ static void *mlx5_ib_add_slave_port(struct mlx5_core_dev *mdev)
 
 static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 {
+	const struct mlx5_ib_profile *profile;
 	enum rdma_link_layer ll;
 	struct mlx5_ib_dev *dev;
 	int port_type_cap;
@@ -6933,7 +6929,12 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	dev->mdev = mdev;
 	dev->num_ports = num_ports;
 
-	return __mlx5_ib_add(dev, &pf_profile);
+	if (ll == IB_LINK_LAYER_ETHERNET && !mlx5_is_roce_enabled(mdev))
+		profile = &raw_eth_profile;
+	else
+		profile = &pf_profile;
+
+	return __mlx5_ib_add(dev, profile);
 }
 
 static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context)

drivers/net/ethernet/mellanox/mlx5/core/devlink.c

@@ -85,6 +85,22 @@ mlx5_devlink_info_get(struct devlink *devlink, struct devlink_info_req *req,
 	return 0;
 }
 
+static int mlx5_devlink_reload_down(struct devlink *devlink, bool netns_change,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	return mlx5_unload_one(dev, false);
+}
+
+static int mlx5_devlink_reload_up(struct devlink *devlink,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	return mlx5_load_one(dev, false);
+}
+
 static const struct devlink_ops mlx5_devlink_ops = {
 #ifdef CONFIG_MLX5_ESWITCH
 	.eswitch_mode_set = mlx5_devlink_eswitch_mode_set,
@@ -96,6 +112,8 @@ static const struct devlink_ops mlx5_devlink_ops = {
 #endif
 	.flash_update = mlx5_devlink_flash_update,
 	.info_get = mlx5_devlink_info_get,
+	.reload_down = mlx5_devlink_reload_down,
+	.reload_up = mlx5_devlink_reload_up,
 };
 
 struct devlink *mlx5_devlink_alloc(void)
@@ -177,12 +195,29 @@ enum mlx5_devlink_param_id {
 	MLX5_DEVLINK_PARAM_FLOW_STEERING_MODE,
 };
 
+static int mlx5_devlink_enable_roce_validate(struct devlink *devlink, u32 id,
+					     union devlink_param_value val,
+					     struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	bool new_state = val.vbool;
+
+	if (new_state && !MLX5_CAP_GEN(dev, roce)) {
+		NL_SET_ERR_MSG_MOD(extack, "Device doesn't support RoCE");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
 static const struct devlink_param mlx5_devlink_params[] = {
 	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_FLOW_STEERING_MODE,
 			     "flow_steering_mode", DEVLINK_PARAM_TYPE_STRING,
 			     BIT(DEVLINK_PARAM_CMODE_RUNTIME),
 			     mlx5_devlink_fs_mode_get, mlx5_devlink_fs_mode_set,
 			     mlx5_devlink_fs_mode_validate),
+	DEVLINK_PARAM_GENERIC(ENABLE_ROCE, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, mlx5_devlink_enable_roce_validate),
 };
 
 static void mlx5_devlink_set_params_init_values(struct devlink *devlink)
@@ -197,6 +232,11 @@ static void mlx5_devlink_set_params_init_values(struct devlink *devlink)
 	devlink_param_driverinit_value_set(devlink,
 					   MLX5_DEVLINK_PARAM_FLOW_STEERING_MODE,
 					   value);
+
+	value.vbool = MLX5_CAP_GEN(dev, roce);
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_ENABLE_ROCE,
+					   value);
 }
 
 int mlx5_devlink_register(struct devlink *devlink, struct device *dev)
@@ -213,6 +253,7 @@ int mlx5_devlink_register(struct devlink *devlink, struct device *dev)
 		goto params_reg_err;
 	mlx5_devlink_set_params_init_values(devlink);
 	devlink_params_publish(devlink);
+	devlink_reload_enable(devlink);
 	return 0;
 
 params_reg_err:
@@ -222,6 +263,7 @@ int mlx5_devlink_register(struct devlink *devlink, struct device *dev)
 
 void mlx5_devlink_unregister(struct devlink *devlink)
 {
+	devlink_reload_disable(devlink);
 	devlink_params_unregister(devlink, mlx5_devlink_params,
 				  ARRAY_SIZE(mlx5_devlink_params));
 	devlink_unregister(devlink);

drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c

@@ -77,8 +77,8 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
 				   struct neighbour **out_n,
 				   u8 *out_ttl)
 {
+	struct neighbour *n;
 	struct rtable *rt;
-	struct neighbour *n = NULL;
 
 #if IS_ENABLED(CONFIG_INET)
 	struct mlx5_core_dev *mdev = priv->mdev;
@@ -138,8 +138,8 @@ static int mlx5e_route_lookup_ipv6(struct mlx5e_priv *priv,
 				   struct neighbour **out_n,
 				   u8 *out_ttl)
 {
-	struct neighbour *n = NULL;
 	struct dst_entry *dst;
+	struct neighbour *n;
 
 #if IS_ENABLED(CONFIG_INET) && IS_ENABLED(CONFIG_IPV6)
 	int ret;
@@ -212,8 +212,8 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv,
 	int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
 	const struct ip_tunnel_key *tun_key = &e->tun_info->key;
 	struct net_device *out_dev, *route_dev;
-	struct neighbour *n = NULL;
 	struct flowi4 fl4 = {};
+	struct neighbour *n;
 	int ipv4_encap_size;
 	char *encap_header;
 	u8 nud_state, ttl;
@@ -239,12 +239,15 @@ int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv,
 	if (max_encap_size < ipv4_encap_size) {
 		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
 			       ipv4_encap_size, max_encap_size);
-		return -EOPNOTSUPP;
+		err = -EOPNOTSUPP;
+		goto out;
 	}
 
 	encap_header = kzalloc(ipv4_encap_size, GFP_KERNEL);
-	if (!encap_header)
-		return -ENOMEM;
+	if (!encap_header) {
+		err = -ENOMEM;
+		goto out;
+	}
 
 	/* used by mlx5e_detach_encap to lookup a neigh hash table
 	 * entry in the neigh hash table when a user deletes a rule
@@ -328,9 +331,9 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv,
 	int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
 	const struct ip_tunnel_key *tun_key = &e->tun_info->key;
 	struct net_device *out_dev, *route_dev;
-	struct neighbour *n = NULL;
 	struct flowi6 fl6 = {};
 	struct ipv6hdr *ip6h;
+	struct neighbour *n;
 	int ipv6_encap_size;
 	char *encap_header;
 	u8 nud_state, ttl;
@@ -355,12 +358,15 @@ int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv,
 	if (max_encap_size < ipv6_encap_size) {
 		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
 			       ipv6_encap_size, max_encap_size);
-		return -EOPNOTSUPP;
+		err = -EOPNOTSUPP;
+		goto out;
 	}
 
 	encap_header = kzalloc(ipv6_encap_size, GFP_KERNEL);
-	if (!encap_header)
-		return -ENOMEM;
+	if (!encap_header) {
+		err = -ENOMEM;
+		goto out;
+	}
 
 	/* used by mlx5e_detach_encap to lookup a neigh hash table
 	 * entry in the neigh hash table when a user deletes a rule

drivers/net/ethernet/mellanox/mlx5/core/en_main.c

@@ -63,6 +63,7 @@
 #include "en/xsk/rx.h"
 #include "en/xsk/tx.h"
 #include "en/hv_vhca_stats.h"
+#include "lib/mlx5.h"
 
 
 bool mlx5e_check_fragmented_striding_rq_cap(struct mlx5_core_dev *mdev)
@@ -5427,6 +5428,7 @@ static void *mlx5e_add(struct mlx5_core_dev *mdev)
 		return NULL;
 	}
 
+	dev_net_set(netdev, mlx5_core_net(mdev));
 	priv = netdev_priv(netdev);
 
 	err = mlx5e_attach(mdev, priv);