Merge tag 'mlx5-updates-2018-02-23' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux

Saeed Mahameed says:

mlx5-update-2018-02-23 (IB representors)

From: Mark Bloch <markb@mellanox.com>
=========
Add IB representor when in switchdev mode

The following series adds support for an IB (RAW Ethernet only) device
representor which is created when the user switches to switchdev mode.

Today when switching to switchdev mode the only representors which are
created are net devices. Each netdev is a representor of a virtual
function and any data sent via the representor is received on the virtual
function, and any data sent via the virtual function is received by the
representor.

For the mlx5 driver the main use of this functionality is to be able to
use Open vSwitch on the hypervisor in order to manage/control traffic
from/to the virtual functions. Open vSwitch can also work with  DPDK
devices and not just net devices, this series exposes an IB device, which
Mellanox PMD driver uses, which then can be used by Open vSwitch DPDK.

An IB device representor exposes only RAW Ethernet QP capabilities and
the ability to create flow rules to direct traffic to its RX queues. The
state of the IB device (ACTIVE/DOWN etc..) is based on the state of the
corresponding net device representor. No other RDMA/RoCE functionality is
currently supported and no GID table is exposed.
=========
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/infiniband/hw/mlx5/Makefile

@@ -2,3 +2,4 @@ obj-$(CONFIG_MLX5_INFINIBAND)	+= mlx5_ib.o
 
 mlx5_ib-y :=	main.o cq.o doorbell.o qp.o mem.o srq.o mr.o ah.o mad.o gsi.o ib_virt.o cmd.o cong.o
 mlx5_ib-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += odp.o
+mlx5_ib-$(CONFIG_MLX5_ESWITCH) += ib_rep.o

drivers/infiniband/hw/mlx5/ib_rep.c

+/* SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause) */
+/*
+ * Copyright (c) 2018 Mellanox Technologies. All rights reserved.
+ */
+
+#include "ib_rep.h"
+
+static const struct mlx5_ib_profile rep_profile = {
+	STAGE_CREATE(MLX5_IB_STAGE_INIT,
+		     mlx5_ib_stage_init_init,
+		     mlx5_ib_stage_init_cleanup),
+	STAGE_CREATE(MLX5_IB_STAGE_FLOW_DB,
+		     mlx5_ib_stage_rep_flow_db_init,
+		     NULL),
+	STAGE_CREATE(MLX5_IB_STAGE_CAPS,
+		     mlx5_ib_stage_caps_init,
+		     NULL),
+	STAGE_CREATE(MLX5_IB_STAGE_NON_DEFAULT_CB,
+		     mlx5_ib_stage_rep_non_default_cb,
+		     NULL),
+	STAGE_CREATE(MLX5_IB_STAGE_ROCE,
+		     mlx5_ib_stage_rep_roce_init,
+		     mlx5_ib_stage_rep_roce_cleanup),
+	STAGE_CREATE(MLX5_IB_STAGE_DEVICE_RESOURCES,
+		     mlx5_ib_stage_dev_res_init,
+		     mlx5_ib_stage_dev_res_cleanup),
+	STAGE_CREATE(MLX5_IB_STAGE_COUNTERS,
+		     mlx5_ib_stage_counters_init,
+		     mlx5_ib_stage_counters_cleanup),
+	STAGE_CREATE(MLX5_IB_STAGE_BFREG,
+		     mlx5_ib_stage_bfrag_init,
+		     mlx5_ib_stage_bfrag_cleanup),
+	STAGE_CREATE(MLX5_IB_STAGE_IB_REG,
+		     mlx5_ib_stage_ib_reg_init,
+		     mlx5_ib_stage_ib_reg_cleanup),
+	STAGE_CREATE(MLX5_IB_STAGE_UMR_RESOURCES,
+		     mlx5_ib_stage_umr_res_init,
+		     mlx5_ib_stage_umr_res_cleanup),
+	STAGE_CREATE(MLX5_IB_STAGE_CLASS_ATTR,
+		     mlx5_ib_stage_class_attr_init,
+		     NULL),
+};
+
+static int
+mlx5_ib_nic_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
+{
+	return 0;
+}
+
+static void
+mlx5_ib_nic_rep_unload(struct mlx5_eswitch_rep *rep)
+{
+	rep->rep_if[REP_IB].priv = NULL;
+}
+
+static int
+mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5_ib_dev *ibdev;
+
+	ibdev = (struct mlx5_ib_dev *)ib_alloc_device(sizeof(*ibdev));
+	if (!ibdev)
+		return -ENOMEM;
+
+	ibdev->rep = rep;
+	ibdev->mdev = dev;
+	ibdev->num_ports = max(MLX5_CAP_GEN(dev, num_ports),
+			       MLX5_CAP_GEN(dev, num_vhca_ports));
+	if (!__mlx5_ib_add(ibdev, &rep_profile))
+		return -EINVAL;
+
+	rep->rep_if[REP_IB].priv = ibdev;
+
+	return 0;
+}
+
+static void
+mlx5_ib_vport_rep_unload(struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5_ib_dev *dev;
+
+	if (!rep->rep_if[REP_IB].priv)
+		return;
+
+	dev = mlx5_ib_rep_to_dev(rep);
+	__mlx5_ib_remove(dev, dev->profile, MLX5_IB_STAGE_MAX);
+	rep->rep_if[REP_IB].priv = NULL;
+}
+
+static void *mlx5_ib_vport_get_proto_dev(struct mlx5_eswitch_rep *rep)
+{
+	return mlx5_ib_rep_to_dev(rep);
+}
+
+static void mlx5_ib_rep_register_vf_vports(struct mlx5_ib_dev *dev)
+{
+	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
+	int total_vfs = MLX5_TOTAL_VPORTS(dev->mdev);
+	int vport;
+
+	for (vport = 1; vport < total_vfs; vport++) {
+		struct mlx5_eswitch_rep_if rep_if = {};
+
+		rep_if.load = mlx5_ib_vport_rep_load;
+		rep_if.unload = mlx5_ib_vport_rep_unload;
+		rep_if.get_proto_dev = mlx5_ib_vport_get_proto_dev;
+		mlx5_eswitch_register_vport_rep(esw, vport, &rep_if, REP_IB);
+	}
+}
+
+static void mlx5_ib_rep_unregister_vf_vports(struct mlx5_ib_dev *dev)
+{
+	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
+	int total_vfs = MLX5_TOTAL_VPORTS(dev->mdev);
+	int vport;
+
+	for (vport = 1; vport < total_vfs; vport++)
+		mlx5_eswitch_unregister_vport_rep(esw, vport, REP_IB);
+}
+
+void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev)
+{
+	struct mlx5_eswitch *esw = dev->mdev->priv.eswitch;
+	struct mlx5_eswitch_rep_if rep_if = {};
+
+	rep_if.load = mlx5_ib_nic_rep_load;
+	rep_if.unload = mlx5_ib_nic_rep_unload;
+	rep_if.get_proto_dev = mlx5_ib_vport_get_proto_dev;
+	rep_if.priv = dev;
+
+	mlx5_eswitch_register_vport_rep(esw, 0, &rep_if, REP_IB);
+
+	mlx5_ib_rep_register_vf_vports(dev);
+}
+
+void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev)
+{
+	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
+
+	mlx5_ib_rep_unregister_vf_vports(dev); /* VFs vports */
+	mlx5_eswitch_unregister_vport_rep(esw, 0, REP_IB); /* UPLINK PF*/
+}
+
+u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw)
+{
+	return mlx5_eswitch_mode(esw);
+}
+
+struct mlx5_ib_dev *mlx5_ib_get_rep_ibdev(struct mlx5_eswitch *esw,
+					  int vport_index)
+{
+	return mlx5_eswitch_get_proto_dev(esw, vport_index, REP_IB);
+}
+
+struct net_device *mlx5_ib_get_rep_netdev(struct mlx5_eswitch *esw,
+					  int vport_index)
+{
+	return mlx5_eswitch_get_proto_dev(esw, vport_index, REP_ETH);
+}
+
+struct mlx5_ib_dev *mlx5_ib_get_uplink_ibdev(struct mlx5_eswitch *esw)
+{
+	return mlx5_eswitch_uplink_get_proto_dev(esw, REP_IB);
+}
+
+struct mlx5_eswitch_rep *mlx5_ib_vport_rep(struct mlx5_eswitch *esw, int vport)
+{
+	return mlx5_eswitch_vport_rep(esw, vport);
+}
+
+int create_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
+			      struct mlx5_ib_sq *sq)
+{
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_eswitch *esw = dev->mdev->priv.eswitch;
+
+	if (!dev->rep)
+		return 0;
+
+	flow_rule =
+		mlx5_eswitch_add_send_to_vport_rule(esw,
+						    dev->rep->vport,
+						    sq->base.mqp.qpn);
+	if (IS_ERR(flow_rule))
+		return PTR_ERR(flow_rule);
+	sq->flow_rule = flow_rule;
+
+	return 0;
+}

drivers/infiniband/hw/mlx5/ib_rep.h

+/* SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause) */
+/*
+ * Copyright (c) 2018 Mellanox Technologies. All rights reserved.
+ */
+
+#ifndef __MLX5_IB_REP_H__
+#define __MLX5_IB_REP_H__
+
+#include <linux/mlx5/eswitch.h>
+#include "mlx5_ib.h"
+
+#ifdef CONFIG_MLX5_ESWITCH
+u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw);
+struct mlx5_ib_dev *mlx5_ib_get_rep_ibdev(struct mlx5_eswitch *esw,
+					  int vport_index);
+struct mlx5_ib_dev *mlx5_ib_get_uplink_ibdev(struct mlx5_eswitch *esw);
+struct mlx5_eswitch_rep *mlx5_ib_vport_rep(struct mlx5_eswitch *esw,
+					   int vport_index);
+void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev);
+void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev);
+int create_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
+			      struct mlx5_ib_sq *sq);
+struct net_device *mlx5_ib_get_rep_netdev(struct mlx5_eswitch *esw,
+					  int vport_index);
+#else /* CONFIG_MLX5_ESWITCH */
+static inline u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw)
+{
+	return SRIOV_NONE;
+}
+
+static inline
+struct mlx5_ib_dev *mlx5_ib_get_rep_ibdev(struct mlx5_eswitch *esw,
+					  int vport_index)
+{
+	return NULL;
+}
+
+static inline
+struct mlx5_ib_dev *mlx5_ib_get_uplink_ibdev(struct mlx5_eswitch *esw)
+{
+	return NULL;
+}
+
+static inline
+struct mlx5_eswitch_rep *mlx5_ib_vport_rep(struct mlx5_eswitch *esw,
+					   int vport_index)
+{
+	return NULL;
+}
+
+static inline void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev) {}
+static inline void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev) {}
+static inline int create_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
+					    struct mlx5_ib_sq *sq)
+{
+	return 0;
+}
+
+static inline
+struct net_device *mlx5_ib_get_rep_netdev(struct mlx5_eswitch *esw,
+					  int vport_index)
+{
+	return NULL;
+}
+#endif
+
+static inline
+struct mlx5_ib_dev *mlx5_ib_rep_to_dev(struct mlx5_eswitch_rep *rep)
+{
+	return (struct mlx5_ib_dev *)rep->rep_if[REP_IB].priv;
+}
+#endif /* __MLX5_IB_REP_H__ */

drivers/infiniband/hw/mlx5/mlx5_ib.h

@@ -343,6 +343,7 @@ struct mlx5_ib_sq {
 	struct mlx5_ib_wq	*sq;
 	struct mlx5_ib_ubuffer  ubuffer;
 	struct mlx5_db		*doorbell;
+	struct mlx5_flow_handle	*flow_rule;
 	u32			tisn;
 	u8			state;
 };
@@ -731,7 +732,9 @@ struct mlx5_ib_delay_drop {
 
 enum mlx5_ib_stages {
 	MLX5_IB_STAGE_INIT,
+	MLX5_IB_STAGE_FLOW_DB,
 	MLX5_IB_STAGE_CAPS,
+	MLX5_IB_STAGE_NON_DEFAULT_CB,
 	MLX5_IB_STAGE_ROCE,
 	MLX5_IB_STAGE_DEVICE_RESOURCES,
 	MLX5_IB_STAGE_ODP,
@@ -743,6 +746,7 @@ enum mlx5_ib_stages {
 	MLX5_IB_STAGE_UMR_RESOURCES,
 	MLX5_IB_STAGE_DELAY_DROP,
 	MLX5_IB_STAGE_CLASS_ATTR,
+	MLX5_IB_STAGE_REP_REG,
 	MLX5_IB_STAGE_MAX,
 };
 
@@ -797,7 +801,7 @@ struct mlx5_ib_dev {
 	struct srcu_struct      mr_srcu;
 	u32			null_mkey;
 #endif
-	struct mlx5_ib_flow_db	flow_db;
+	struct mlx5_ib_flow_db	*flow_db;
 	/* protect resources needed as part of reset flow */
 	spinlock_t		reset_flow_resource_lock;
 	struct list_head	qp_list;
@@ -807,6 +811,7 @@ struct mlx5_ib_dev {
 	struct mlx5_sq_bfreg	fp_bfreg;
 	struct mlx5_ib_delay_drop	delay_drop;
 	const struct mlx5_ib_profile	*profile;
+	struct mlx5_eswitch_rep		*rep;
 
 	/* protect the user_td */
 	struct mutex		lb_mutex;
@@ -1049,6 +1054,31 @@ static inline void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset,
 
 #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 
+/* Needed for rep profile */
+int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev);
+void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_rep_flow_db_init(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_rep_non_default_cb(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_rep_roce_init(struct mlx5_ib_dev *dev);
+void mlx5_ib_stage_rep_roce_cleanup(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_dev_res_init(struct mlx5_ib_dev *dev);
+void mlx5_ib_stage_dev_res_cleanup(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_counters_init(struct mlx5_ib_dev *dev);
+void mlx5_ib_stage_counters_cleanup(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_bfrag_init(struct mlx5_ib_dev *dev);
+void mlx5_ib_stage_bfrag_cleanup(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_ib_reg_init(struct mlx5_ib_dev *dev);
+void mlx5_ib_stage_ib_reg_cleanup(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_umr_res_init(struct mlx5_ib_dev *dev);
+void mlx5_ib_stage_umr_res_cleanup(struct mlx5_ib_dev *dev);
+int mlx5_ib_stage_class_attr_init(struct mlx5_ib_dev *dev);
+void __mlx5_ib_remove(struct mlx5_ib_dev *dev,
+		      const struct mlx5_ib_profile *profile,
+		      int stage);
+void *__mlx5_ib_add(struct mlx5_ib_dev *dev,
+		    const struct mlx5_ib_profile *profile);
+
 int mlx5_ib_get_vf_config(struct ib_device *device, int vf,
 			  u8 port, struct ifla_vf_info *info);
 int mlx5_ib_set_vf_link_state(struct ib_device *device, int vf,

drivers/infiniband/hw/mlx5/mr.c

@@ -587,7 +587,7 @@ static void clean_keys(struct mlx5_ib_dev *dev, int c)
 
 static void mlx5_mr_cache_debugfs_cleanup(struct mlx5_ib_dev *dev)
 {
-	if (!mlx5_debugfs_root)
+	if (!mlx5_debugfs_root || dev->rep)
 		return;
 
 	debugfs_remove_recursive(dev->cache.root);
@@ -600,7 +600,7 @@ static int mlx5_mr_cache_debugfs_init(struct mlx5_ib_dev *dev)
 	struct mlx5_cache_ent *ent;
 	int i;
 
-	if (!mlx5_debugfs_root)
+	if (!mlx5_debugfs_root || dev->rep)
 		return 0;
 
 	cache->root = debugfs_create_dir("mr_cache", dev->mdev->priv.dbg_root);
@@ -690,6 +690,7 @@ int mlx5_mr_cache_init(struct mlx5_ib_dev *dev)
 			   MLX5_IB_UMR_OCTOWORD;
 		ent->access_mode = MLX5_MKC_ACCESS_MODE_MTT;
 		if ((dev->mdev->profile->mask & MLX5_PROF_MASK_MR_CACHE) &&
+		    !dev->rep &&
 		    mlx5_core_is_pf(dev->mdev))
 			ent->limit = dev->mdev->profile->mr_cache[i].limit;
 		else

drivers/infiniband/hw/mlx5/qp.c

@@ -36,6 +36,7 @@
 #include <rdma/ib_user_verbs.h>
 #include <linux/mlx5/fs.h>
 #include "mlx5_ib.h"
+#include "ib_rep.h"
 
 /* not supported currently */
 static int wq_signature;
@@ -1082,6 +1083,13 @@ static void destroy_raw_packet_qp_tis(struct mlx5_ib_dev *dev,
 	mlx5_core_destroy_tis(dev->mdev, sq->tisn);
 }
 
+static void destroy_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
+				       struct mlx5_ib_sq *sq)
+{
+	if (sq->flow_rule)
+		mlx5_del_flow_rules(sq->flow_rule);
+}
+
 static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev,
 				   struct mlx5_ib_sq *sq, void *qpin,
 				   struct ib_pd *pd)
@@ -1145,8 +1153,15 @@ static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev,
 	if (err)
 		goto err_umem;
 
+	err = create_flow_rule_vport_sq(dev, sq);
+	if (err)
+		goto err_flow;
+
 	return 0;
 
+err_flow:
+	mlx5_core_destroy_sq_tracked(dev->mdev, &sq->base.mqp);
+
 err_umem:
 	ib_umem_release(sq->ubuffer.umem);
 	sq->ubuffer.umem = NULL;
@@ -1157,6 +1172,7 @@ static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev,
 static void destroy_raw_packet_qp_sq(struct mlx5_ib_dev *dev,
 				     struct mlx5_ib_sq *sq)
 {
+	destroy_flow_rule_vport_sq(dev, sq);
 	mlx5_core_destroy_sq_tracked(dev->mdev, &sq->base.mqp);
 	ib_umem_release(sq->ubuffer.umem);
 }
@@ -1263,6 +1279,10 @@ static int create_raw_packet_qp_tir(struct mlx5_ib_dev *dev,
 	if (tunnel_offload_en)
 		MLX5_SET(tirc, tirc, tunneled_offload_en, 1);
 
+	if (dev->rep)
+		MLX5_SET(tirc, tirc, self_lb_block,
+			 MLX5_TIRC_SELF_LB_BLOCK_BLOCK_UNICAST_);
+
 	err = mlx5_core_create_tir(dev->mdev, in, inlen, &rq->tirn);
 
 	kvfree(in);
@@ -1554,6 +1574,10 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
 	MLX5_SET(rx_hash_field_select, hfso, selected_fields, selected_fields);
 
 create_tir:
+	if (dev->rep)
+		MLX5_SET(tirc, tirc, self_lb_block,
+			 MLX5_TIRC_SELF_LB_BLOCK_BLOCK_UNICAST_);
+
 	err = mlx5_core_create_tir(dev->mdev, in, inlen, &qp->rss_qp.tirn);
 
 	if (err)

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

@@ -337,6 +337,14 @@ void mlx5_unregister_interface(struct mlx5_interface *intf)
 }
 EXPORT_SYMBOL(mlx5_unregister_interface);
 
+void mlx5_reload_interface(struct mlx5_core_dev *mdev, int protocol)
+{
+	mutex_lock(&mlx5_intf_mutex);
+	mlx5_remove_dev_by_protocol(mdev, protocol);
+	mlx5_add_dev_by_protocol(mdev, protocol);
+	mutex_unlock(&mlx5_intf_mutex);
+}
+
 void *mlx5_get_protocol_dev(struct mlx5_core_dev *mdev, int protocol)
 {
 	struct mlx5_priv *priv = &mdev->priv;

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

@@ -1156,6 +1156,15 @@ mlx5e_vport_rep_unload(struct mlx5_eswitch_rep *rep)
 	kfree(ppriv); /* mlx5e_rep_priv */
 }
 
+static void *mlx5e_vport_rep_get_proto_dev(struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_rep_priv *rpriv;
+
+	rpriv = mlx5e_rep_to_rep_priv(rep);
+
+	return rpriv->netdev;
+}
+
 static void mlx5e_rep_register_vf_vports(struct mlx5e_priv *priv)
 {
 	struct mlx5_core_dev *mdev = priv->mdev;
@@ -1168,6 +1177,7 @@ static void mlx5e_rep_register_vf_vports(struct mlx5e_priv *priv)
 
 		rep_if.load = mlx5e_vport_rep_load;
 		rep_if.unload = mlx5e_vport_rep_unload;
+		rep_if.get_proto_dev = mlx5e_vport_rep_get_proto_dev;
 		mlx5_eswitch_register_vport_rep(esw, vport, &rep_if, REP_ETH);
 	}
 }
@@ -1195,6 +1205,7 @@ void mlx5e_register_vport_reps(struct mlx5e_priv *priv)
 
 	rep_if.load = mlx5e_nic_rep_load;
 	rep_if.unload = mlx5e_nic_rep_unload;
+	rep_if.get_proto_dev = mlx5e_vport_rep_get_proto_dev;
 	rep_if.priv = rpriv;
 	INIT_LIST_HEAD(&rpriv->vport_sqs_list);
 	mlx5_eswitch_register_vport_rep(esw, 0, &rep_if, REP_ETH); /* UPLINK PF vport*/

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

@@ -1619,10 +1619,14 @@ int mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs, int mode)
 	esw_info(esw->dev, "E-Switch enable SRIOV: nvfs(%d) mode (%d)\n", nvfs, mode);
 	esw->mode = mode;
 
-	if (mode == SRIOV_LEGACY)
+	if (mode == SRIOV_LEGACY) {
 		err = esw_create_legacy_fdb_table(esw, nvfs + 1);
-	else
+	} else {
+		mlx5_reload_interface(esw->dev, MLX5_INTERFACE_PROTOCOL_IB);
+
 		err = esw_offloads_init(esw, nvfs + 1);
+	}
+
 	if (err)
 		goto abort;
 
@@ -1644,12 +1648,17 @@ int mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs, int mode)
 
 abort:
 	esw->mode = SRIOV_NONE;
+
+	if (mode == SRIOV_OFFLOADS)
+		mlx5_reload_interface(esw->dev, MLX5_INTERFACE_PROTOCOL_IB);
+
 	return err;
 }
 
 void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw)
 {
 	struct esw_mc_addr *mc_promisc;
+	int old_mode;
 	int nvports;
 	int i;
 
@@ -1675,7 +1684,11 @@ void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw)
 	else if (esw->mode == SRIOV_OFFLOADS)
 		esw_offloads_cleanup(esw, nvports);
 
+	old_mode = esw->mode;
 	esw->mode = SRIOV_NONE;
+
+	if (old_mode == SRIOV_OFFLOADS)
+		mlx5_reload_interface(esw->dev, MLX5_INTERFACE_PROTOCOL_IB);
 }
 
 int mlx5_eswitch_init(struct mlx5_core_dev *dev)
@@ -2175,3 +2188,9 @@ int mlx5_eswitch_get_vport_stats(struct mlx5_eswitch *esw,
 	kvfree(out);
 	return err;
 }
+
+u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw)
+{
+	return esw->mode;
+}
+EXPORT_SYMBOL_GPL(mlx5_eswitch_mode);

drivers/net/ethernet/mellanox/mlx5/core/eswitch.h

@@ -37,19 +37,9 @@
 #include <linux/if_link.h>
 #include <net/devlink.h>
 #include <linux/mlx5/device.h>
+#include <linux/mlx5/eswitch.h>
 #include "lib/mpfs.h"
 
-enum {
-	SRIOV_NONE,
-	SRIOV_LEGACY,
-	SRIOV_OFFLOADS
-};
-
-enum {
-	REP_ETH,
-	NUM_REP_TYPES,
-};
-
 #ifdef CONFIG_MLX5_ESWITCH
 
 #define MLX5_MAX_UC_PER_VPORT(dev) \
@@ -139,29 +129,13 @@ struct mlx5_eswitch_fdb {
 			struct mlx5_flow_table *fdb;
 			struct mlx5_flow_group *send_to_vport_grp;
 			struct mlx5_flow_group *miss_grp;
-			struct mlx5_flow_handle *miss_rule;
+			struct mlx5_flow_handle *miss_rule_uni;
+			struct mlx5_flow_handle *miss_rule_multi;
 			int vlan_push_pop_refcount;
 		} offloads;
 	};
 };
 
-struct mlx5_eswitch_rep;
-struct mlx5_eswitch_rep_if {
-	int		       (*load)(struct mlx5_core_dev *dev,
-				       struct mlx5_eswitch_rep *rep);
-	void		       (*unload)(struct mlx5_eswitch_rep *rep);
-	void			*priv;
-	bool		       valid;
-};
-
-struct mlx5_eswitch_rep {
-	struct mlx5_eswitch_rep_if rep_if[NUM_REP_TYPES];
-	u16		       vport;
-	u8		       hw_id[ETH_ALEN];
-	u16		       vlan;
-	u32		       vlan_refcount;
-};
-
 struct mlx5_esw_offload {
 	struct mlx5_flow_table *ft_offloads;
 	struct mlx5_flow_group *vport_rx_group;
@@ -231,9 +205,6 @@ int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw,
 int mlx5_eswitch_get_vport_stats(struct mlx5_eswitch *esw,
 				 int vport,
 				 struct ifla_vf_stats *vf_stats);
-struct mlx5_flow_handle *
-mlx5_eswitch_add_send_to_vport_rule(struct mlx5_eswitch *esw, int vport,
-				    u32 sqn);
 void mlx5_eswitch_del_send_to_vport_rule(struct mlx5_flow_handle *rule);
 
 struct mlx5_flow_spec;
@@ -278,13 +249,6 @@ int mlx5_devlink_eswitch_inline_mode_get(struct devlink *devlink, u8 *mode);
 int mlx5_eswitch_inline_mode_get(struct mlx5_eswitch *esw, int nvfs, u8 *mode);
 int mlx5_devlink_eswitch_encap_mode_set(struct devlink *devlink, u8 encap);
 int mlx5_devlink_eswitch_encap_mode_get(struct devlink *devlink, u8 *encap);
-void mlx5_eswitch_register_vport_rep(struct mlx5_eswitch *esw,
-				     int vport_index,
-				     struct mlx5_eswitch_rep_if *rep_if,
-				     u8 rep_type);
-void mlx5_eswitch_unregister_vport_rep(struct mlx5_eswitch *esw,
-				       int vport_index,
-				       u8 rep_type);
 void *mlx5_eswitch_get_uplink_priv(struct mlx5_eswitch *esw, u8 rep_type);
 
 int mlx5_eswitch_add_vlan_action(struct mlx5_eswitch *esw,

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

@@ -338,6 +338,7 @@ mlx5_eswitch_add_send_to_vport_rule(struct mlx5_eswitch *esw, int vport, u32 sqn
 	kvfree(spec);
 	return flow_rule;
 }
+EXPORT_SYMBOL(mlx5_eswitch_add_send_to_vport_rule);
 
 void mlx5_eswitch_del_send_to_vport_rule(struct mlx5_flow_handle *rule)
 {
@@ -350,7 +351,11 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 	struct mlx5_flow_destination dest = {};
 	struct mlx5_flow_handle *flow_rule = NULL;
 	struct mlx5_flow_spec *spec;
+	void *headers_c;
+	void *headers_v;
 	int err = 0;
+	u8 *dmac_c;
+	u8 *dmac_v;
 
 	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
 	if (!spec) {
@@ -358,6 +363,13 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 		goto out;
 	}
 
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				 outer_headers);
+	dmac_c = MLX5_ADDR_OF(fte_match_param, headers_c,
+			      outer_headers.dmac_47_16);
+	dmac_c[0] = 0x01;
+
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
 	dest.vport_num = 0;
 	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
@@ -366,11 +378,28 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 					&flow_act, &dest, 1);
 	if (IS_ERR(flow_rule)) {
 		err = PTR_ERR(flow_rule);
-		esw_warn(esw->dev,  "FDB: Failed to add miss flow rule err %d\n", err);
+		esw_warn(esw->dev,  "FDB: Failed to add unicast miss flow rule err %d\n", err);
 		goto out;
 	}
 
-	esw->fdb_table.offloads.miss_rule = flow_rule;
+	esw->fdb_table.offloads.miss_rule_uni = flow_rule;
+
+	headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				 outer_headers);
+	dmac_v = MLX5_ADDR_OF(fte_match_param, headers_v,
+			      outer_headers.dmac_47_16);
+	dmac_v[0] = 0x01;
+	flow_rule = mlx5_add_flow_rules(esw->fdb_table.offloads.fdb, spec,
+					&flow_act, &dest, 1);
+	if (IS_ERR(flow_rule)) {
+		err = PTR_ERR(flow_rule);
+		esw_warn(esw->dev, "FDB: Failed to add multicast miss flow rule err %d\n", err);
+		mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_uni);
+		goto out;
+	}
+
+	esw->fdb_table.offloads.miss_rule_multi = flow_rule;
+
 out:
 	kvfree(spec);
 	return err;
@@ -426,6 +455,7 @@ static void esw_destroy_offloads_fast_fdb_table(struct mlx5_eswitch *esw)
 }
 
 #define MAX_PF_SQ 256
+#define MAX_SQ_NVPORTS 32
 
 static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 {
@@ -438,6 +468,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 	struct mlx5_flow_group *g;
 	void *match_criteria;
 	u32 *flow_group_in;
+	u8 *dmac;
 
 	esw_debug(esw->dev, "Create offloads FDB Tables\n");
 	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
@@ -455,7 +486,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 	if (err)
 		goto fast_fdb_err;
 
-	table_size = nvports + MAX_PF_SQ + 1;
+	table_size = nvports * MAX_SQ_NVPORTS + MAX_PF_SQ + 2;
 
 	ft_attr.max_fte = table_size;
 	ft_attr.prio = FDB_SLOW_PATH;
@@ -478,7 +509,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 	MLX5_SET_TO_ONES(fte_match_param, match_criteria, misc_parameters.source_sqn);
 	MLX5_SET_TO_ONES(fte_match_param, match_criteria, misc_parameters.source_port);
 
-	ix = nvports + MAX_PF_SQ;
+	ix = nvports * MAX_SQ_NVPORTS + MAX_PF_SQ;
 	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
 	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, ix - 1);
 
@@ -492,10 +523,16 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 
 	/* create miss group */
 	memset(flow_group_in, 0, inlen);
-	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS);
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+				      match_criteria);
+	dmac = MLX5_ADDR_OF(fte_match_param, match_criteria,
+			    outer_headers.dmac_47_16);
+	dmac[0] = 0x01;
 
 	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, ix);
-	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, ix + 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, ix + 2);
 
 	g = mlx5_create_flow_group(fdb, flow_group_in);
 	if (IS_ERR(g)) {
@@ -531,7 +568,8 @@ static void esw_destroy_offloads_fdb_tables(struct mlx5_eswitch *esw)
 		return;
 
 	esw_debug(esw->dev, "Destroy offloads FDB Tables\n");
-	mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule);
+	mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_multi);
+	mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_uni);
 	mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_grp);
 	mlx5_destroy_flow_group(esw->fdb_table.offloads.miss_grp);
 
@@ -789,14 +827,9 @@ int esw_offloads_init(struct mlx5_eswitch *esw, int nvports)
 {
 	int err;
 
-	/* disable PF RoCE so missed packets don't go through RoCE steering */
-	mlx5_dev_list_lock();
-	mlx5_remove_dev_by_protocol(esw->dev, MLX5_INTERFACE_PROTOCOL_IB);
-	mlx5_dev_list_unlock();
-
 	err = esw_create_offloads_fdb_tables(esw, nvports);
 	if (err)
-		goto create_fdb_err;
+		return err;
 
 	err = esw_create_offloads_table(esw);
 	if (err)
@@ -821,12 +854,6 @@ int esw_offloads_init(struct mlx5_eswitch *esw, int nvports)
 create_ft_err:
 	esw_destroy_offloads_fdb_tables(esw);
 
-create_fdb_err:
-	/* enable back PF RoCE */
-	mlx5_dev_list_lock();
-	mlx5_add_dev_by_protocol(esw->dev, MLX5_INTERFACE_PROTOCOL_IB);
-	mlx5_dev_list_unlock();
-
 	return err;
 }
 
@@ -844,9 +871,7 @@ static int esw_offloads_stop(struct mlx5_eswitch *esw)
 	}
 
 	/* enable back PF RoCE */
-	mlx5_dev_list_lock();
-	mlx5_add_dev_by_protocol(esw->dev, MLX5_INTERFACE_PROTOCOL_IB);
-	mlx5_dev_list_unlock();
+	mlx5_reload_interface(esw->dev, MLX5_INTERFACE_PROTOCOL_IB);
 
 	return err;
 }
@@ -1160,10 +1185,12 @@ void mlx5_eswitch_register_vport_rep(struct mlx5_eswitch *esw,
 
 	rep_if->load   = __rep_if->load;
 	rep_if->unload = __rep_if->unload;
+	rep_if->get_proto_dev = __rep_if->get_proto_dev;
 	rep_if->priv = __rep_if->priv;
 
 	rep_if->valid = true;
 }
+EXPORT_SYMBOL(mlx5_eswitch_register_vport_rep);
 
 void mlx5_eswitch_unregister_vport_rep(struct mlx5_eswitch *esw,
 				       int vport_index, u8 rep_type)
@@ -1178,6 +1205,7 @@ void mlx5_eswitch_unregister_vport_rep(struct mlx5_eswitch *esw,
 
 	rep->rep_if[rep_type].valid = false;
 }
+EXPORT_SYMBOL(mlx5_eswitch_unregister_vport_rep);
 
 void *mlx5_eswitch_get_uplink_priv(struct mlx5_eswitch *esw, u8 rep_type)
 {
@@ -1188,3 +1216,35 @@ void *mlx5_eswitch_get_uplink_priv(struct mlx5_eswitch *esw, u8 rep_type)
 	rep = &offloads->vport_reps[UPLINK_REP_INDEX];
 	return rep->rep_if[rep_type].priv;
 }
+
+void *mlx5_eswitch_get_proto_dev(struct mlx5_eswitch *esw,
+				 int vport,
+				 u8 rep_type)
+{
+	struct mlx5_esw_offload *offloads = &esw->offloads;
+	struct mlx5_eswitch_rep *rep;
+
+	if (vport == FDB_UPLINK_VPORT)
+		vport = UPLINK_REP_INDEX;
+
+	rep = &offloads->vport_reps[vport];
+
+	if (rep->rep_if[rep_type].valid &&
+	    rep->rep_if[rep_type].get_proto_dev)
+		return rep->rep_if[rep_type].get_proto_dev(rep);
+	return NULL;
+}
+EXPORT_SYMBOL(mlx5_eswitch_get_proto_dev);
+
+void *mlx5_eswitch_uplink_get_proto_dev(struct mlx5_eswitch *esw, u8 rep_type)
+{
+	return mlx5_eswitch_get_proto_dev(esw, UPLINK_REP_INDEX, rep_type);
+}
+EXPORT_SYMBOL(mlx5_eswitch_uplink_get_proto_dev);
+
+struct mlx5_eswitch_rep *mlx5_eswitch_vport_rep(struct mlx5_eswitch *esw,
+						int vport)
+{
+	return &esw->offloads.vport_reps[vport];
+}
+EXPORT_SYMBOL(mlx5_eswitch_vport_rep);

drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h

@@ -43,12 +43,6 @@
 #define DRIVER_NAME "mlx5_core"
 #define DRIVER_VERSION "5.0-0"
 
-#define MLX5_TOTAL_VPORTS(mdev) (1 + pci_sriov_get_totalvfs(mdev->pdev))
-#define MLX5_VPORT_MANAGER(mdev) \
-	(MLX5_CAP_GEN(mdev, vport_group_manager) && \
-	(MLX5_CAP_GEN(mdev, port_type) == MLX5_CAP_PORT_TYPE_ETH) && \
-	 mlx5_core_is_pf(mdev))
-
 extern uint mlx5_core_debug_mask;
 
 #define mlx5_core_dbg(__dev, format, ...)				\
@@ -207,4 +201,5 @@ static inline int mlx5_lag_is_lacp_owner(struct mlx5_core_dev *dev)
 int mlx5_lag_allow(struct mlx5_core_dev *dev);
 int mlx5_lag_forbid(struct mlx5_core_dev *dev);
 
+void mlx5_reload_interface(struct mlx5_core_dev *mdev, int protocol);
 #endif /* __MLX5_CORE_H__ */

include/linux/mlx5/driver.h

@@ -1224,6 +1224,12 @@ static inline int mlx5_core_is_pf(struct mlx5_core_dev *dev)
 	return !(dev->priv.pci_dev_data & MLX5_PCI_DEV_IS_VF);
 }
 
+#define MLX5_TOTAL_VPORTS(mdev) (1 + pci_sriov_get_totalvfs((mdev)->pdev))
+#define MLX5_VPORT_MANAGER(mdev) \
+	(MLX5_CAP_GEN(mdev, vport_group_manager) && \
+	 (MLX5_CAP_GEN(mdev, port_type) == MLX5_CAP_PORT_TYPE_ETH) && \
+	 mlx5_core_is_pf(mdev))
+
 static inline int mlx5_get_gid_table_len(u16 param)
 {
 	if (param > 4) {

include/linux/mlx5/eswitch.h

+/* SPDX-License-Identifier: (GPL-2.0+ OR BSD-3-Clause) */
+/*
+ * Copyright (c) 2018 Mellanox Technologies. All rights reserved.
+ */
+
+#ifndef _MLX5_ESWITCH_
+#define _MLX5_ESWITCH_
+
+#include <linux/mlx5/driver.h>
+
+enum {
+	SRIOV_NONE,
+	SRIOV_LEGACY,
+	SRIOV_OFFLOADS
+};
+
+enum {
+	REP_ETH,
+	REP_IB,
+	NUM_REP_TYPES,
+};
+
+struct mlx5_eswitch_rep;
+struct mlx5_eswitch_rep_if {
+	int		       (*load)(struct mlx5_core_dev *dev,
+				       struct mlx5_eswitch_rep *rep);
+	void		       (*unload)(struct mlx5_eswitch_rep *rep);
+	void		       *(*get_proto_dev)(struct mlx5_eswitch_rep *rep);
+	void			*priv;
+	bool		       valid;
+};
+
+struct mlx5_eswitch_rep {
+	struct mlx5_eswitch_rep_if rep_if[NUM_REP_TYPES];
+	u16		       vport;
+	u8		       hw_id[ETH_ALEN];
+	u16		       vlan;
+	u32		       vlan_refcount;
+};
+
+void mlx5_eswitch_register_vport_rep(struct mlx5_eswitch *esw,
+				     int vport_index,
+				     struct mlx5_eswitch_rep_if *rep_if,
+				     u8 rep_type);
+void mlx5_eswitch_unregister_vport_rep(struct mlx5_eswitch *esw,
+				       int vport_index,
+				       u8 rep_type);
+void *mlx5_eswitch_get_proto_dev(struct mlx5_eswitch *esw,
+				 int vport,
+				 u8 rep_type);
+struct mlx5_eswitch_rep *mlx5_eswitch_vport_rep(struct mlx5_eswitch *esw,
+						int vport);
+void *mlx5_eswitch_uplink_get_proto_dev(struct mlx5_eswitch *esw, u8 rep_type);
+u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw);
+struct mlx5_flow_handle *
+mlx5_eswitch_add_send_to_vport_rule(struct mlx5_eswitch *esw,
+				    int vport, u32 sqn);
+#endif