5 files changed, 204 insertions, 36 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
index 6e7001c0cfd4..15e41dc84d53 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
@@ -5,12 +5,13 @@
 
 static char *get_str_mode_type(struct mlx5_lag *ldev)
 {
-	if (ldev->mode == MLX5_LAG_MODE_ROCE)
-		return "roce";
-	if (ldev->mode == MLX5_LAG_MODE_SRIOV)
-		return "switchdev";
-	if (ldev->mode == MLX5_LAG_MODE_MULTIPATH)
-		return "multipath";
+	switch (ldev->mode) {
+	case MLX5_LAG_MODE_ROCE: return "roce";
+	case MLX5_LAG_MODE_SRIOV: return "switchdev";
+	case MLX5_LAG_MODE_MULTIPATH: return "multipath";
+	case MLX5_LAG_MODE_MPESW: return "multiport_eswitch";
+	default: return "invalid";
+	}
 
 	return NULL;
 }
@@ -43,11 +44,11 @@ static int port_sel_mode_show(struct seq_file *file, void *priv)
 	ldev = dev->priv.lag;
 	mutex_lock(&ldev->lock);
 	if (__mlx5_lag_is_active(ldev))
-		mode = get_str_port_sel_mode(ldev->mode_flags);
+		mode = mlx5_get_str_port_sel_mode(ldev);
 	else
 		ret = -EINVAL;
 	mutex_unlock(&ldev->lock);
-	if (ret || !mode)
+	if (ret)
 		return ret;
 
 	seq_printf(file, "%s\n", mode);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
index 5c3900586d23..552b6e26e701 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
@@ -41,6 +41,7 @@
 #include "esw/acl/ofld.h"
 #include "lag.h"
 #include "mp.h"
+#include "mpesw.h"
 
 enum {
 	MLX5_LAG_EGRESS_PORT_1 = 1,
@@ -58,6 +59,9 @@ static int get_port_sel_mode(enum mlx5_lag_mode mode, unsigned long flags)
 	if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags))
 		return MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_FT;
 
+	if (mode == MLX5_LAG_MODE_MPESW)
+		return MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_MPESW;
+
 	return MLX5_LAG_PORT_SELECT_MODE_QUEUE_AFFINITY;
 }
 
@@ -196,7 +200,8 @@ static void mlx5_ldev_free(struct kref *ref)
 	if (ldev->nb.notifier_call)
 		unregister_netdevice_notifier_net(&init_net, &ldev->nb);
 	mlx5_lag_mp_cleanup(ldev);
-	cancel_delayed_work_sync(&ldev->bond_work);
+	mlx5_lag_mpesw_cleanup(ldev);
+	cancel_work_sync(&ldev->mpesw_work);
 	destroy_workqueue(ldev->wq);
 	mutex_destroy(&ldev->lock);
 	kfree(ldev);
@@ -242,6 +247,8 @@ static struct mlx5_lag *mlx5_lag_dev_alloc(struct mlx5_core_dev *dev)
 	if (err)
 		mlx5_core_err(dev, "Failed to init multipath lag err=%d\n",
 			      err);
+
+	mlx5_lag_mpesw_init(ldev);
 	ldev->ports = MLX5_CAP_GEN(dev, num_lag_ports);
 	ldev->buckets = 1;
 
@@ -442,16 +449,19 @@ static int mlx5_lag_set_port_sel_mode_roce(struct mlx5_lag *ldev,
 	return 0;
 }
 
-static int mlx5_lag_set_port_sel_mode_offloads(struct mlx5_lag *ldev,
-					       struct lag_tracker *tracker, unsigned long *flags)
+static void mlx5_lag_set_port_sel_mode_offloads(struct mlx5_lag *ldev,
+						struct lag_tracker *tracker,
+						enum mlx5_lag_mode mode,
+						unsigned long *flags)
 {
 	struct lag_func *dev0 = &ldev->pf[MLX5_LAG_P1];
 
+	if (mode == MLX5_LAG_MODE_MPESW)
+		return;
+
 	if (MLX5_CAP_PORT_SELECTION(dev0->dev, port_select_flow_table) &&
 	    tracker->tx_type == NETDEV_LAG_TX_TYPE_HASH)
 		set_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, flags);
-
-	return 0;
 }
 
 static int mlx5_lag_set_flags(struct mlx5_lag *ldev, enum mlx5_lag_mode mode,
@@ -467,14 +477,20 @@ static int mlx5_lag_set_flags(struct mlx5_lag *ldev, enum mlx5_lag_mode mode,
 	if (roce_lag)
 		return mlx5_lag_set_port_sel_mode_roce(ldev, flags);
 
-	return mlx5_lag_set_port_sel_mode_offloads(ldev, tracker, flags);
+	mlx5_lag_set_port_sel_mode_offloads(ldev, tracker, mode, flags);
+	return 0;
 }
 
-char *get_str_port_sel_mode(unsigned long flags)
+char *mlx5_get_str_port_sel_mode(struct mlx5_lag *ldev)
 {
-	if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags))
-		return "hash";
-	return "queue_affinity";
+	int port_sel_mode = get_port_sel_mode(ldev->mode, ldev->mode_flags);
+
+	switch (port_sel_mode) {
+	case MLX5_LAG_PORT_SELECT_MODE_QUEUE_AFFINITY: return "queue_affinity";
+	case MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_FT: return "hash";
+	case MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_MPESW: return "mpesw";
+	default: return "invalid";
+	}
 }
 
 static int mlx5_create_lag(struct mlx5_lag *ldev,
@@ -488,9 +504,10 @@ static int mlx5_create_lag(struct mlx5_lag *ldev,
 	u32 in[MLX5_ST_SZ_DW(destroy_lag_in)] = {};
 	int err;
 
-	mlx5_lag_print_mapping(dev0, ldev, tracker, flags);
+	if (tracker)
+		mlx5_lag_print_mapping(dev0, ldev, tracker, flags);
 	mlx5_core_info(dev0, "shared_fdb:%d mode:%s\n",
-		       shared_fdb, get_str_port_sel_mode(flags));
+		       shared_fdb, mlx5_get_str_port_sel_mode(ldev));
 
 	err = mlx5_cmd_create_lag(dev0, ldev->v2p_map, mode, flags);
 	if (err) {
@@ -526,22 +543,24 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
 {
 	bool roce_lag = mode == MLX5_LAG_MODE_ROCE;
 	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
-	unsigned long flags;
+	unsigned long flags = 0;
 	int err;
 
 	err = mlx5_lag_set_flags(ldev, mode, tracker, shared_fdb, &flags);
 	if (err)
 		return err;
 
-	mlx5_infer_tx_affinity_mapping(tracker, ldev->ports, ldev->buckets, ldev->v2p_map);
-	if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags)) {
-		err = mlx5_lag_port_sel_create(ldev, tracker->hash_type,
-					       ldev->v2p_map);
-		if (err) {
-			mlx5_core_err(dev0,
-				      "Failed to create LAG port selection(%d)\n",
-				      err);
-			return err;
+	if (mode != MLX5_LAG_MODE_MPESW) {
+		mlx5_infer_tx_affinity_mapping(tracker, ldev->ports, ldev->buckets, ldev->v2p_map);
+		if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags)) {
+			err = mlx5_lag_port_sel_create(ldev, tracker->hash_type,
+						       ldev->v2p_map);
+			if (err) {
+				mlx5_core_err(dev0,
+					      "Failed to create LAG port selection(%d)\n",
+					      err);
+				return err;
+			}
 		}
 	}
 
@@ -559,7 +578,7 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
 		return err;
 	}
 
-	if (tracker->tx_type == NETDEV_LAG_TX_TYPE_ACTIVEBACKUP &&
+	if (tracker && tracker->tx_type == NETDEV_LAG_TX_TYPE_ACTIVEBACKUP &&
 	    !roce_lag)
 		mlx5_lag_drop_rule_setup(ldev, tracker);
 
@@ -675,7 +694,7 @@ static void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
 	}
 }
 
-static void mlx5_disable_lag(struct mlx5_lag *ldev)
+void mlx5_disable_lag(struct mlx5_lag *ldev)
 {
 	bool shared_fdb = test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &ldev->mode_flags);
 	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
@@ -712,7 +731,7 @@ static void mlx5_disable_lag(struct mlx5_lag *ldev)
 	}
 }
 
-static bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev)
+bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev)
 {
 	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
 	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
@@ -748,6 +767,18 @@ static bool mlx5_lag_is_roce_lag(struct mlx5_lag *ldev)
 	return roce_lag;
 }
 
+static bool mlx5_lag_should_modify_lag(struct mlx5_lag *ldev, bool do_bond)
+{
+	return do_bond && __mlx5_lag_is_active(ldev) &&
+	       ldev->mode != MLX5_LAG_MODE_MPESW;
+}
+
+static bool mlx5_lag_should_disable_lag(struct mlx5_lag *ldev, bool do_bond)
+{
+	return !do_bond && __mlx5_lag_is_active(ldev) &&
+	       ldev->mode != MLX5_LAG_MODE_MPESW;
+}
+
 static void mlx5_do_bond(struct mlx5_lag *ldev)
 {
 	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
@@ -810,9 +841,9 @@ static void mlx5_do_bond(struct mlx5_lag *ldev)
 				return;
 			}
 		}
-	} else if (do_bond && __mlx5_lag_is_active(ldev)) {
+	} else if (mlx5_lag_should_modify_lag(ldev, do_bond)) {
 		mlx5_modify_lag(ldev, &tracker);
-	} else if (!do_bond && __mlx5_lag_is_active(ldev)) {
+	} else if (mlx5_lag_should_disable_lag(ldev, do_bond)) {
 		mlx5_disable_lag(ldev);
 	}
 }
@@ -986,6 +1017,7 @@ static int mlx5_handle_changeinfodata_event(struct mlx5_lag *ldev,
 	return 1;
 }
 
+/* this handler is always registered to netdev events */
 static int mlx5_lag_netdev_event(struct notifier_block *this,
 				 unsigned long event, void *ptr)
 {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
index 244b548e1420..72f70fad4641 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
@@ -10,6 +10,7 @@
 #include "mlx5_core.h"
 #include "mp.h"
 #include "port_sel.h"
+#include "mpesw.h"
 
 enum {
 	MLX5_LAG_P1,
@@ -30,6 +31,7 @@ enum mlx5_lag_mode {
 	MLX5_LAG_MODE_ROCE,
 	MLX5_LAG_MODE_SRIOV,
 	MLX5_LAG_MODE_MULTIPATH,
+	MLX5_LAG_MODE_MPESW,
 };
 
 struct lag_func {
@@ -63,11 +65,13 @@ struct mlx5_lag {
 	struct lag_tracker        tracker;
 	struct workqueue_struct   *wq;
 	struct delayed_work       bond_work;
+	struct work_struct	  mpesw_work;
 	struct notifier_block     nb;
 	struct lag_mp             lag_mp;
 	struct mlx5_lag_port_sel  port_sel;
 	/* Protect lag fields/state changes */
 	struct mutex		  lock;
+	struct lag_mpesw	  lag_mpesw;
 };
 
 static inline struct mlx5_lag *
@@ -96,12 +100,16 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
 		      bool shared_fdb);
 int mlx5_lag_dev_get_netdev_idx(struct mlx5_lag *ldev,
 				struct net_device *ndev);
+bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev);
+void mlx5_lag_del_mpesw_rule(struct mlx5_core_dev *dev);
+int mlx5_lag_add_mpesw_rule(struct mlx5_core_dev *dev);
 
-char *get_str_port_sel_mode(unsigned long flags);
+char *mlx5_get_str_port_sel_mode(struct mlx5_lag *ldev);
 void mlx5_infer_tx_enabled(struct lag_tracker *tracker, u8 num_ports,
 			   u8 *ports, int *num_enabled);
 
 void mlx5_ldev_add_debugfs(struct mlx5_core_dev *dev);
 void mlx5_ldev_remove_debugfs(struct dentry *dbg);
+void mlx5_disable_lag(struct mlx5_lag *ldev);
 
 #endif /* __MLX5_LAG_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
new file mode 100644
index 000000000000..ee4b25a50315
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -0,0 +1,101 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/netdevice.h>
+#include <net/nexthop.h>
+#include "lag/lag.h"
+#include "eswitch.h"
+#include "lib/mlx5.h"
+
+void mlx5_mpesw_work(struct work_struct *work)
+{
+	struct mlx5_lag *ldev = container_of(work, struct mlx5_lag, mpesw_work);
+
+	mutex_lock(&ldev->lock);
+	mlx5_disable_lag(ldev);
+	mutex_unlock(&ldev->lock);
+}
+
+static void mlx5_lag_disable_mpesw(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev = dev->priv.lag;
+
+	if (!queue_work(ldev->wq, &ldev->mpesw_work))
+		mlx5_core_warn(dev, "failed to queue work\n");
+}
+
+void mlx5_lag_del_mpesw_rule(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev = dev->priv.lag;
+
+	if (!ldev)
+		return;
+
+	mutex_lock(&ldev->lock);
+	if (!atomic_dec_return(&ldev->lag_mpesw.mpesw_rule_count) &&
+	    ldev->mode == MLX5_LAG_MODE_MPESW)
+		mlx5_lag_disable_mpesw(dev);
+	mutex_unlock(&ldev->lock);
+}
+
+int mlx5_lag_add_mpesw_rule(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev = dev->priv.lag;
+	bool shared_fdb;
+	int err = 0;
+
+	if (!ldev)
+		return 0;
+
+	mutex_lock(&ldev->lock);
+	if (atomic_add_return(1, &ldev->lag_mpesw.mpesw_rule_count) != 1)
+		goto out;
+
+	if (ldev->mode != MLX5_LAG_MODE_NONE) {
+		err = -EINVAL;
+		goto out;
+	}
+	shared_fdb = mlx5_shared_fdb_supported(ldev);
+	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, shared_fdb);
+	if (err)
+		mlx5_core_warn(dev, "Failed to create LAG in MPESW mode (%d)\n", err);
+
+out:
+	mutex_unlock(&ldev->lock);
+	return err;
+}
+
+int mlx5_lag_do_mirred(struct mlx5_core_dev *mdev, struct net_device *out_dev)
+{
+	struct mlx5_lag *ldev = mdev->priv.lag;
+
+	if (!netif_is_bond_master(out_dev) || !ldev)
+		return 0;
+
+	mutex_lock(&ldev->lock);
+	if (ldev->mode == MLX5_LAG_MODE_MPESW) {
+		mutex_unlock(&ldev->lock);
+		return -EOPNOTSUPP;
+	}
+	mutex_unlock(&ldev->lock);
+	return 0;
+}
+
+bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev)
+{
+	bool ret;
+
+	ret = dev->priv.lag && dev->priv.lag->mode == MLX5_LAG_MODE_MPESW;
+	return ret;
+}
+
+void mlx5_lag_mpesw_init(struct mlx5_lag *ldev)
+{
+	INIT_WORK(&ldev->mpesw_work, mlx5_mpesw_work);
+	atomic_set(&ldev->lag_mpesw.mpesw_rule_count, 0);
+}
+
+void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev)
+{
+	cancel_delayed_work_sync(&ldev->bond_work);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
new file mode 100644
index 000000000000..d39a02280e29
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
@@ -0,0 +1,26 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_LAG_MPESW_H__
+#define __MLX5_LAG_MPESW_H__
+
+#include "lag.h"
+#include "mlx5_core.h"
+
+struct lag_mpesw {
+	struct work_struct mpesw_work;
+	atomic_t mpesw_rule_count;
+};
+
+void mlx5_mpesw_work(struct work_struct *work);
+int mlx5_lag_do_mirred(struct mlx5_core_dev *mdev, struct net_device *out_dev);
+bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev);
+#if IS_ENABLED(CONFIG_MLX5_ESWITCH)
+void mlx5_lag_mpesw_init(struct mlx5_lag *ldev);
+void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev);
+#else
+void mlx5_lag_mpesw_init(struct mlx5_lag *ldev) {}
+void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev) {}
+#endif
+
+#endif /* __MLX5_LAG_MPESW_H__ */