Merge tag 'mlx5-updates-2023-06-09' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux

mlx5-updates-2023-06-09

1) Embedded CPU Virtual Functions
2) Lightweight local SFs

Daniel Jurgens says:
====================
Embedded CPU Virtual Functions

This series enables the creation of virtual functions on Bluefield (the
embedded CPU platform). Embedded CPU virtual functions (EC VFs). EC VF
creation, deletion and management interfaces are the same as those for
virtual functions in a server with a Connect-X NIC.

When using EC VFs on the ARM the creation of virtual functions on the
host system is still supported. Host VFs eswitch vports occupy a range
of 1..max_vfs, the EC VF vport range is max_vfs+1..max_ec_vfs.

Every function (PF, ECPF, VF, EC VF, and subfunction) has a function ID
associated with it. Prior to this series the function ID and the eswitch
vport were the same. That is no longer the case, the EC VF function ID
range is 1..max_ec_vfs. When querying or setting the capabilities of an
EC VF function an new bit must be set in the query/set HCA cap
structure.

This is a high level overview of the changes made:
- Allocate vports for EC VFs if they are enabled.
- Create representors and devlink ports for the EC VF vports.
- When querying/setting HCA caps by vport break the assumption
that function ID is the same a vport number and adjust
accordingly.
- Create a new type of page, so that when SRIOV on the ARM is
disabled, but remains enabled on the host, the driver can
wait for the correct pages.
- Update SRIOV code to support EC VF creation/deletion.

===================

Lightweight local SFs:

Last 3 patches form Shay Drory:

SFs are heavy weight and by default they come with the full package of
ConnectX features. Usually users want specialized SFs for one specific
purpose and using devlink users will almost always override the set of
advertises features of an SF and reload it.

Shay Drory says:
================
In order to avoid the wasted time and resources on the reload, local SFs
will probe without any auxiliary sub-device, so that the SFs can be
configured prior to its full probe.

The defaults of the enable_* devlink params of these SFs are set to
false.

Usage example:
Create SF:
$ devlink port add pci/0000:08:00.0 flavour pcisf pfnum 0 sfnum 11
$ devlink port function set pci/0000:08:00.0/32768 \
hw_addr 00:00:00:00:00:11 state active

Enable ETH auxiliary device:
$ devlink dev param set auxiliary/mlx5_core.sf.1 \
name enable_eth value true cmode driverinit

Now, in order to fully probe the SF, use devlink reload:
$ devlink dev reload auxiliary/mlx5_core.sf.1

At this point the user have SF devlink instance with auxiliary device
for the Ethernet functionality only.

================

Documentation/networking/device_drivers/ethernet/mellanox/mlx5/switchdev.rst

···
 Subfunction
 ===========
 
+Subfunction which are spawned over the E-switch are created only with devlink
+device, and by default all the SF auxiliary devices are disabled.
+This will allow user to configure the SF before the SF have been fully probed,
+which will save time.
+
+Usage example:
+Create SF:
+$ devlink port add pci/0000:08:00.0 flavour pcisf pfnum 0 sfnum 11
+$ devlink port function set pci/0000:08:00.0/32768 \
+               hw_addr 00:00:00:00:00:11 state active
+
+Enable ETH auxiliary device:
+$ devlink dev param set auxiliary/mlx5_core.sf.1 \
+              name enable_eth value true cmode driverinit
+
+Now, in order to fully probe the SF, use devlink reload:
+$ devlink dev reload auxiliary/mlx5_core.sf.1
+
+mlx5 supports ETH,rdma and vdpa (vnet) auxiliary devices devlink params (see :ref:`Documentation/networking/devlink/devlink-params.rst`)
+
 mlx5 supports subfunction management using devlink port (see :ref:`Documentation/networking/devlink/devlink-port.rst <devlink_port>`) interface.
 
 A subfunction has its own function capabilities and its own resources. This

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

···
 
 	debugfs_create_u32("fw_pages_total", 0400, pages, &dev->priv.fw_pages);
 	debugfs_create_u32("fw_pages_vfs", 0400, pages, &dev->priv.page_counters[MLX5_VF]);
+	debugfs_create_u32("fw_pages_ec_vfs", 0400, pages, &dev->priv.page_counters[MLX5_EC_VF]);
 	debugfs_create_u32("fw_pages_sfs", 0400, pages, &dev->priv.page_counters[MLX5_SF]);
 	debugfs_create_u32("fw_pages_host_pf", 0400, pages, &dev->priv.page_counters[MLX5_HOST_PF]);
 	debugfs_create_u32("fw_pages_alloc_failed", 0400, pages, &dev->priv.fw_pages_alloc_failed);

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

···
 	auxiliary_device_uninit(adev);
 }
 
+void mlx5_dev_set_lightweight(struct mlx5_core_dev *dev)
+{
+	mutex_lock(&mlx5_intf_mutex);
+	dev->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV;
+	mutex_unlock(&mlx5_intf_mutex);
+}
+
+bool mlx5_dev_is_lightweight(struct mlx5_core_dev *dev)
+{
+	return dev->priv.flags & MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV;
+}
+
 int mlx5_attach_device(struct mlx5_core_dev *dev)
 {
 	struct mlx5_priv *priv = &dev->priv;
···
 		bool is_supported = false;
 
 		if (priv->adev[i])
+			continue;
+
+		if (mlx5_adev_devices[i].is_enabled &&
+		    !(mlx5_adev_devices[i].is_enabled(dev)))
 			continue;
 
 		if (mlx5_adev_devices[i].is_supported)

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

···
 #include "fw_reset.h"
 #include "fs_core.h"
 #include "eswitch.h"
-#include "lag/lag.h"
 #include "esw/qos.h"
 #include "sf/dev/dev.h"
 #include "sf/sf.h"
···
 	bool sf_dev_allocated;
 	int ret = 0;
 
+	if (mlx5_dev_is_lightweight(dev)) {
+		if (action != DEVLINK_RELOAD_ACTION_DRIVER_REINIT)
+			return -EOPNOTSUPP;
+		mlx5_unload_one_light(dev);
+		return 0;
+	}
+
 	sf_dev_allocated = mlx5_sf_dev_allocated(dev);
 	if (sf_dev_allocated) {
 		/* Reload results in deleting SF device which further results in
···
 	*actions_performed = BIT(action);
 	switch (action) {
 	case DEVLINK_RELOAD_ACTION_DRIVER_REINIT:
+		if (mlx5_dev_is_lightweight(dev)) {
+			mlx5_fw_reporters_create(dev);
+			return mlx5_init_one_devl_locked(dev);
+		}
 		ret = mlx5_load_one_devl_locked(dev, false);
 		break;
 	case DEVLINK_RELOAD_ACTION_FW_ACTIVATE:
···
 
 	return 0;
 }
-
-static int mlx5_devlink_esw_multiport_set(struct devlink *devlink, u32 id,
-					  struct devlink_param_gset_ctx *ctx)
-{
-	struct mlx5_core_dev *dev = devlink_priv(devlink);
-
-	if (!MLX5_ESWITCH_MANAGER(dev))
-		return -EOPNOTSUPP;
-
-	if (ctx->val.vbool)
-		return mlx5_lag_mpesw_enable(dev);
-
-	mlx5_lag_mpesw_disable(dev);
-	return 0;
-}
-
-static int mlx5_devlink_esw_multiport_get(struct devlink *devlink, u32 id,
-					  struct devlink_param_gset_ctx *ctx)
-{
-	struct mlx5_core_dev *dev = devlink_priv(devlink);
-
-	if (!MLX5_ESWITCH_MANAGER(dev))
-		return -EOPNOTSUPP;
-
-	ctx->val.vbool = mlx5_lag_is_mpesw(dev);
-	return 0;
-}
 #endif
 
 static int mlx5_devlink_eq_depth_validate(struct devlink *devlink, u32 id,
···
 			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
 			     NULL, NULL,
 			     mlx5_devlink_large_group_num_validate),
-	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_ID_ESW_MULTIPORT,
-			     "esw_multiport", DEVLINK_PARAM_TYPE_BOOL,
-			     BIT(DEVLINK_PARAM_CMODE_RUNTIME),
-			     mlx5_devlink_esw_multiport_get,
-			     mlx5_devlink_esw_multiport_set,
-			     NULL),
 #endif
 	DEVLINK_PARAM_GENERIC(IO_EQ_SIZE, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
 			      NULL, NULL, mlx5_devlink_eq_depth_validate),
···
 	struct mlx5_core_dev *dev = devlink_priv(devlink);
 	union devlink_param_value value;
 
-	value.vbool = MLX5_CAP_GEN(dev, roce);
+	value.vbool = MLX5_CAP_GEN(dev, roce) && !mlx5_dev_is_lightweight(dev);
 	devl_param_driverinit_value_set(devlink,
 					DEVLINK_PARAM_GENERIC_ID_ENABLE_ROCE,
 					value);
···
 	if (err)
 		return err;
 
-	value.vbool = true;
+	value.vbool = !mlx5_dev_is_lightweight(dev);
 	devl_param_driverinit_value_set(devlink,
 					DEVLINK_PARAM_GENERIC_ID_ENABLE_ETH,
 					value);
···
 
 static int mlx5_devlink_rdma_params_register(struct devlink *devlink)
 {
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
 	union devlink_param_value value;
 	int err;
 
···
 	if (err)
 		return err;
 
-	value.vbool = true;
+	value.vbool = !mlx5_dev_is_lightweight(dev);
 	devl_param_driverinit_value_set(devlink,
 					DEVLINK_PARAM_GENERIC_ID_ENABLE_RDMA,
 					value);
···
 	if (err)
 		return err;
 
-	value.vbool = true;
+	value.vbool = !mlx5_dev_is_lightweight(dev);
 	devl_param_driverinit_value_set(devlink,
 					DEVLINK_PARAM_GENERIC_ID_ENABLE_VNET,
 					value);

drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_act.c

···
 	int err;
 
 	handle = kzalloc(sizeof(*handle), GFP_KERNEL);
-	if (!handle) {
-		kfree(handle);
+	if (!handle)
 		return ERR_PTR(-ENOMEM);
-	}
 
 	post_attr->chain = 0;
 	post_attr->prio = 0;

drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c

···
 {
 	return vport_num == MLX5_VPORT_UPLINK ||
 	       (mlx5_core_is_ecpf(esw->dev) && vport_num == MLX5_VPORT_PF) ||
-	       mlx5_eswitch_is_vf_vport(esw, vport_num);
+	       mlx5_eswitch_is_vf_vport(esw, vport_num) ||
+	       mlx5_core_is_ec_vf_vport(esw->dev, vport_num);
 }
 
 static struct devlink_port *mlx5_esw_dl_port_alloc(struct mlx5_eswitch *esw, u16 vport_num)
···
 		dl_port->attrs.switch_id.id_len = ppid.id_len;
 		devlink_port_attrs_pci_vf_set(dl_port, controller_num, pfnum,
 					      vport_num - 1, external);
+	}  else if (mlx5_core_is_ec_vf_vport(esw->dev, vport_num)) {
+		memcpy(dl_port->attrs.switch_id.id, ppid.id, ppid.id_len);
+		dl_port->attrs.switch_id.id_len = ppid.id_len;
+		devlink_port_attrs_pci_vf_set(dl_port, controller_num, pfnum,
+					      vport_num - 1, false);
 	}
 	return dl_port;
 }

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

···
 #include "esw/qos.h"
 #include "mlx5_core.h"
 #include "lib/eq.h"
+#include "lag/lag.h"
 #include "eswitch.h"
 #include "fs_core.h"
 #include "devlink.h"
···
 	}
 }
 
+static void mlx5_eswitch_clear_ec_vf_vports_info(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	mlx5_esw_for_each_ec_vf_vport(esw, i, vport, esw->esw_funcs.num_ec_vfs) {
+		memset(&vport->qos, 0, sizeof(vport->qos));
+		memset(&vport->info, 0, sizeof(vport->info));
+		vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO;
+	}
+}
+
 /* Public E-Switch API */
 int mlx5_eswitch_load_vport(struct mlx5_eswitch *esw, u16 vport_num,
 			    enum mlx5_eswitch_vport_event enabled_events)
···
 	}
 }
 
+static void mlx5_eswitch_unload_ec_vf_vports(struct mlx5_eswitch *esw,
+					     u16 num_ec_vfs)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	mlx5_esw_for_each_ec_vf_vport(esw, i, vport, num_ec_vfs) {
+		if (!vport->enabled)
+			continue;
+		mlx5_eswitch_unload_vport(esw, vport->vport);
+	}
+}
+
 int mlx5_eswitch_load_vf_vports(struct mlx5_eswitch *esw, u16 num_vfs,
 				enum mlx5_eswitch_vport_event enabled_events)
 {
···
 
 vf_err:
 	mlx5_eswitch_unload_vf_vports(esw, num_vfs);
+	return err;
+}
+
+static int mlx5_eswitch_load_ec_vf_vports(struct mlx5_eswitch *esw, u16 num_ec_vfs,
+					  enum mlx5_eswitch_vport_event enabled_events)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+	int err;
+
+	mlx5_esw_for_each_ec_vf_vport(esw, i, vport, num_ec_vfs) {
+		err = mlx5_eswitch_load_vport(esw, vport->vport, enabled_events);
+		if (err)
+			goto vf_err;
+	}
+
+	return 0;
+
+vf_err:
+	mlx5_eswitch_unload_ec_vf_vports(esw, num_ec_vfs);
 	return err;
 }
 
···
 		ret = mlx5_eswitch_load_vport(esw, MLX5_VPORT_ECPF, enabled_events);
 		if (ret)
 			goto ecpf_err;
+		if (mlx5_core_ec_sriov_enabled(esw->dev)) {
+			ret = mlx5_eswitch_load_ec_vf_vports(esw, esw->esw_funcs.num_ec_vfs,
+							     enabled_events);
+			if (ret)
+				goto ec_vf_err;
+		}
 	}
 
 	/* Enable VF vports */
···
 	return 0;
 
 vf_err:
+	if (mlx5_core_ec_sriov_enabled(esw->dev))
+		mlx5_eswitch_unload_ec_vf_vports(esw, esw->esw_funcs.num_ec_vfs);
+ec_vf_err:
 	if (mlx5_ecpf_vport_exists(esw->dev))
 		mlx5_eswitch_unload_vport(esw, MLX5_VPORT_ECPF);
 ecpf_err:
···
 {
 	mlx5_eswitch_unload_vf_vports(esw, esw->esw_funcs.num_vfs);
 
-	if (mlx5_ecpf_vport_exists(esw->dev))
+	if (mlx5_ecpf_vport_exists(esw->dev)) {
+		if (mlx5_core_ec_sriov_enabled(esw->dev))
+			mlx5_eswitch_unload_ec_vf_vports(esw, esw->esw_funcs.num_vfs);
 		mlx5_eswitch_unload_vport(esw, MLX5_VPORT_ECPF);
+	}
 
 	host_pf_disable_hca(esw->dev);
 	mlx5_eswitch_unload_vport(esw, MLX5_VPORT_PF);
···
 
 	esw->esw_funcs.num_vfs = MLX5_GET(query_esw_functions_out, out,
 					  host_params_context.host_num_of_vfs);
+	if (mlx5_core_ec_sriov_enabled(esw->dev))
+		esw->esw_funcs.num_ec_vfs = num_vfs;
+
 	kvfree(out);
 }
 
···
 
 	mlx5_eswitch_event_handlers_register(esw);
 
-	esw_info(esw->dev, "Enable: mode(%s), nvfs(%d), active vports(%d)\n",
+	esw_info(esw->dev, "Enable: mode(%s), nvfs(%d), necvfs(%d), active vports(%d)\n",
 		 esw->mode == MLX5_ESWITCH_LEGACY ? "LEGACY" : "OFFLOADS",
-		 esw->esw_funcs.num_vfs, esw->enabled_vports);
+		 esw->esw_funcs.num_vfs, esw->esw_funcs.num_ec_vfs, esw->enabled_vports);
 
 	mlx5_esw_mode_change_notify(esw, esw->mode);
 
···
 int mlx5_eswitch_enable(struct mlx5_eswitch *esw, int num_vfs)
 {
 	bool toggle_lag;
-	int ret;
+	int ret = 0;
 
 	if (!mlx5_esw_allowed(esw))
 		return 0;
···
 
 		vport_events = (esw->mode == MLX5_ESWITCH_LEGACY) ?
 					MLX5_LEGACY_SRIOV_VPORT_EVENTS : MLX5_VPORT_UC_ADDR_CHANGE;
-		ret = mlx5_eswitch_load_vf_vports(esw, num_vfs, vport_events);
-		if (!ret)
-			esw->esw_funcs.num_vfs = num_vfs;
+		/* If this is the ECPF the number of host VFs is managed via the
+		 * eswitch function change event handler, and any num_vfs provided
+		 * here are intended to be EC VFs.
+		 */
+		if (!mlx5_core_is_ecpf(esw->dev)) {
+			ret = mlx5_eswitch_load_vf_vports(esw, num_vfs, vport_events);
+			if (!ret)
+				esw->esw_funcs.num_vfs = num_vfs;
+		} else if (mlx5_core_ec_sriov_enabled(esw->dev)) {
+			ret = mlx5_eswitch_load_ec_vf_vports(esw, num_vfs, vport_events);
+			if (!ret)
+				esw->esw_funcs.num_ec_vfs = num_vfs;
+		}
 	}
+
 	up_write(&esw->mode_lock);
 
 	if (toggle_lag)
···
 	/* If driver is unloaded, this function is called twice by remove_one()
 	 * and mlx5_unload(). Prevent the second call.
 	 */
-	if (!esw->esw_funcs.num_vfs && !clear_vf)
+	if (!esw->esw_funcs.num_vfs && !esw->esw_funcs.num_ec_vfs && !clear_vf)
 		goto unlock;
 
-	esw_info(esw->dev, "Unload vfs: mode(%s), nvfs(%d), active vports(%d)\n",
+	esw_info(esw->dev, "Unload vfs: mode(%s), nvfs(%d), necvfs(%d), active vports(%d)\n",
 		 esw->mode == MLX5_ESWITCH_LEGACY ? "LEGACY" : "OFFLOADS",
-		 esw->esw_funcs.num_vfs, esw->enabled_vports);
+		 esw->esw_funcs.num_vfs, esw->esw_funcs.num_ec_vfs, esw->enabled_vports);
 
-	mlx5_eswitch_unload_vf_vports(esw, esw->esw_funcs.num_vfs);
-	if (clear_vf)
-		mlx5_eswitch_clear_vf_vports_info(esw);
+	if (!mlx5_core_is_ecpf(esw->dev)) {
+		mlx5_eswitch_unload_vf_vports(esw, esw->esw_funcs.num_vfs);
+		if (clear_vf)
+			mlx5_eswitch_clear_vf_vports_info(esw);
+	} else if (mlx5_core_ec_sriov_enabled(esw->dev)) {
+		mlx5_eswitch_unload_ec_vf_vports(esw, esw->esw_funcs.num_ec_vfs);
+		if (clear_vf)
+			mlx5_eswitch_clear_ec_vf_vports_info(esw);
+	}
 
 	if (esw->mode == MLX5_ESWITCH_OFFLOADS) {
 		struct devlink *devlink = priv_to_devlink(esw->dev);
···
 	if (esw->mode == MLX5_ESWITCH_LEGACY)
 		mlx5_eswitch_disable_locked(esw);
 
-	esw->esw_funcs.num_vfs = 0;
+	if (!mlx5_core_is_ecpf(esw->dev))
+		esw->esw_funcs.num_vfs = 0;
+	else
+		esw->esw_funcs.num_ec_vfs = 0;
 
 unlock:
 	up_write(&esw->mode_lock);
···
 
 	mlx5_eswitch_event_handlers_unregister(esw);
 
-	esw_info(esw->dev, "Disable: mode(%s), nvfs(%d), active vports(%d)\n",
+	esw_info(esw->dev, "Disable: mode(%s), nvfs(%d), necvfs(%d), active vports(%d)\n",
 		 esw->mode == MLX5_ESWITCH_LEGACY ? "LEGACY" : "OFFLOADS",
-		 esw->esw_funcs.num_vfs, esw->enabled_vports);
+		 esw->esw_funcs.num_vfs, esw->esw_funcs.num_ec_vfs, esw->enabled_vports);
 
 	if (esw->fdb_table.flags & MLX5_ESW_FDB_CREATED) {
 		esw->fdb_table.flags &= ~MLX5_ESW_FDB_CREATED;
···
 		idx++;
 	}
 
+	if (mlx5_core_ec_sriov_enabled(esw->dev)) {
+		int ec_vf_base_num = mlx5_core_ec_vf_vport_base(dev);
+
+		for (i = 0; i < mlx5_core_max_ec_vfs(esw->dev); i++) {
+			err = mlx5_esw_vport_alloc(esw, idx, ec_vf_base_num + i);
+			if (err)
+				goto err;
+			idx++;
+		}
+	}
+
 	if (mlx5_ecpf_vport_exists(dev) ||
 	    mlx5_core_is_ecpf_esw_manager(dev)) {
 		err = mlx5_esw_vport_alloc(esw, idx, MLX5_VPORT_ECPF);
···
 	return err;
 }
 
+static int mlx5_devlink_esw_multiport_set(struct devlink *devlink, u32 id,
+					  struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return -EOPNOTSUPP;
+
+	if (ctx->val.vbool)
+		return mlx5_lag_mpesw_enable(dev);
+
+	mlx5_lag_mpesw_disable(dev);
+	return 0;
+}
+
+static int mlx5_devlink_esw_multiport_get(struct devlink *devlink, u32 id,
+					  struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	ctx->val.vbool = mlx5_lag_is_mpesw(dev);
+	return 0;
+}
+
+static const struct devlink_param mlx5_eswitch_params[] = {
+	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_ID_ESW_MULTIPORT,
+			     "esw_multiport", DEVLINK_PARAM_TYPE_BOOL,
+			     BIT(DEVLINK_PARAM_CMODE_RUNTIME),
+			     mlx5_devlink_esw_multiport_get,
+			     mlx5_devlink_esw_multiport_set, NULL),
+};
+
 int mlx5_eswitch_init(struct mlx5_core_dev *dev)
 {
 	struct mlx5_eswitch *esw;
···
 	if (!MLX5_VPORT_MANAGER(dev) && !MLX5_ESWITCH_MANAGER(dev))
 		return 0;
 
+	err = devl_params_register(priv_to_devlink(dev), mlx5_eswitch_params,
+				   ARRAY_SIZE(mlx5_eswitch_params));
+	if (err)
+		return err;
+
 	esw = kzalloc(sizeof(*esw), GFP_KERNEL);
-	if (!esw)
-		return -ENOMEM;
+	if (!esw) {
+		err = -ENOMEM;
+		goto unregister_param;
+	}
 
 	esw->dev = dev;
 	esw->manager_vport = mlx5_eswitch_manager_vport(dev);
···
 	if (esw->work_queue)
 		destroy_workqueue(esw->work_queue);
 	kfree(esw);
+unregister_param:
+	devl_params_unregister(priv_to_devlink(dev), mlx5_eswitch_params,
+			       ARRAY_SIZE(mlx5_eswitch_params));
 	return err;
 }
 
···
 	esw_offloads_cleanup(esw);
 	mlx5_esw_vports_cleanup(esw);
 	kfree(esw);
+	devl_params_unregister(priv_to_devlink(esw->dev), mlx5_eswitch_params,
+			       ARRAY_SIZE(mlx5_eswitch_params));
 }
 
 /* Vport Administration */

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

···
 struct mlx5_esw_functions {
 	struct mlx5_nb		nb;
 	u16			num_vfs;
+	u16			num_ec_vfs;
 };
 
 enum {
···
 
 #define mlx5_esw_for_each_host_func_vport(esw, index, vport, last)	\
 	mlx5_esw_for_each_vport_marked(esw, index, vport, last, MLX5_ESW_VPT_HOST_FN)
+
+/* This macro should only be used if EC SRIOV is enabled.
+ *
+ * Because there were no more marks available on the xarray this uses a
+ * for_each_range approach. The range is only valid when EC SRIOV is enabled
+ */
+#define mlx5_esw_for_each_ec_vf_vport(esw, index, vport, last)		\
+	xa_for_each_range(&((esw)->vports),				\
+			  index,					\
+			  vport,					\
+			  MLX5_CAP_GEN_2((esw->dev), ec_vf_vport_base),	\
+			  (last) - 1)
 
 struct mlx5_eswitch *mlx5_devlink_eswitch_get(struct devlink *devlink);
 struct mlx5_vport *__must_check

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

···
 #define mlx5_esw_for_each_rep(esw, i, rep) \
 	xa_for_each(&((esw)->offloads.vport_reps), i, rep)
 
-#define mlx5_esw_for_each_sf_rep(esw, i, rep) \
-	xa_for_each_marked(&((esw)->offloads.vport_reps), i, rep, MLX5_ESW_VPT_SF)
-
-#define mlx5_esw_for_each_vf_rep(esw, index, rep)	\
-	mlx5_esw_for_each_entry_marked(&((esw)->offloads.vport_reps), index, \
-				       rep, (esw)->esw_funcs.num_vfs, MLX5_ESW_VPT_VF)
-
 /* There are two match-all miss flows, one for unicast dst mac and
  * one for multicast.
  */
···
 		flows[vport->index] = flow;
 	}
 
+	if (mlx5_core_ec_sriov_enabled(esw->dev)) {
+		mlx5_esw_for_each_ec_vf_vport(esw, i, vport, mlx5_core_max_ec_vfs(esw->dev)) {
+			if (i >= mlx5_core_max_ec_vfs(peer_dev))
+				break;
+			esw_set_peer_miss_rule_source_port(esw, peer_dev->priv.eswitch,
+							   spec, vport->vport);
+			flow = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+						   spec, &flow_act, &dest, 1);
+			if (IS_ERR(flow)) {
+				err = PTR_ERR(flow);
+				goto add_ec_vf_flow_err;
+			}
+			flows[vport->index] = flow;
+		}
+	}
 	esw->fdb_table.offloads.peer_miss_rules[mlx5_get_dev_index(peer_dev)] = flows;
 
 	kvfree(spec);
 	return 0;
 
+add_ec_vf_flow_err:
+	mlx5_esw_for_each_ec_vf_vport(esw, i, vport, mlx5_core_max_ec_vfs(esw->dev)) {
+		if (!flows[vport->index])
+			continue;
+		mlx5_del_flow_rules(flows[vport->index]);
+	}
 add_vf_flow_err:
 	mlx5_esw_for_each_vf_vport(esw, i, vport, mlx5_core_max_vfs(esw->dev)) {
 		if (!flows[vport->index])
···
 	unsigned long i;
 
 	flows = esw->fdb_table.offloads.peer_miss_rules[mlx5_get_dev_index(peer_dev)];
+
+	if (mlx5_core_ec_sriov_enabled(esw->dev)) {
+		mlx5_esw_for_each_ec_vf_vport(esw, i, vport, mlx5_core_max_ec_vfs(esw->dev)) {
+			/* The flow for a particular vport could be NULL if the other ECPF
+			 * has fewer or no VFs enabled
+			 */
+			if (!flows[vport->index])
+				continue;
+			mlx5_del_flow_rules(flows[vport->index]);
+		}
+	}
 
 	mlx5_esw_for_each_vf_vport(esw, i, vport, mlx5_core_max_vfs(esw->dev))
 		mlx5_del_flow_rules(flows[vport->index]);
···
 	return 0;
 }
 
-static void mlx5_esw_offloads_rep_mark_set(struct mlx5_eswitch *esw,
-					   struct mlx5_eswitch_rep *rep,
-					   xa_mark_t mark)
-{
-	bool mark_set;
-
-	/* Copy the mark from vport to its rep */
-	mark_set = xa_get_mark(&esw->vports, rep->vport, mark);
-	if (mark_set)
-		xa_set_mark(&esw->offloads.vport_reps, rep->vport, mark);
-}
-
 static int mlx5_esw_offloads_rep_init(struct mlx5_eswitch *esw, const struct mlx5_vport *vport)
 {
 	struct mlx5_eswitch_rep *rep;
···
 	if (err)
 		goto insert_err;
 
-	mlx5_esw_offloads_rep_mark_set(esw, rep, MLX5_ESW_VPT_HOST_FN);
-	mlx5_esw_offloads_rep_mark_set(esw, rep, MLX5_ESW_VPT_VF);
-	mlx5_esw_offloads_rep_mark_set(esw, rep, MLX5_ESW_VPT_SF);
 	return 0;
 
 insert_err:
···
 		esw->offloads.rep_ops[rep_type]->unload(rep);
 }
 
-static void __unload_reps_sf_vport(struct mlx5_eswitch *esw, u8 rep_type)
-{
-	struct mlx5_eswitch_rep *rep;
-	unsigned long i;
-
-	mlx5_esw_for_each_sf_rep(esw, i, rep)
-		__esw_offloads_unload_rep(esw, rep, rep_type);
-}
-
 static void __unload_reps_all_vport(struct mlx5_eswitch *esw, u8 rep_type)
 {
 	struct mlx5_eswitch_rep *rep;
 	unsigned long i;
 
-	__unload_reps_sf_vport(esw, rep_type);
-
-	mlx5_esw_for_each_vf_rep(esw, i, rep)
+	mlx5_esw_for_each_rep(esw, i, rep)
 		__esw_offloads_unload_rep(esw, rep, rep_type);
-
-	if (mlx5_ecpf_vport_exists(esw->dev)) {
-		rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_ECPF);
-		__esw_offloads_unload_rep(esw, rep, rep_type);
-	}
-
-	if (mlx5_core_is_ecpf_esw_manager(esw->dev)) {
-		rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_PF);
-		__esw_offloads_unload_rep(esw, rep, rep_type);
-	}
-
-	rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_UPLINK);
-	__esw_offloads_unload_rep(esw, rep, rep_type);
 }
 
 int mlx5_esw_offloads_rep_load(struct mlx5_eswitch *esw, u16 vport_num)
···
 	/* Representor will control the vport link state */
 	mlx5_esw_for_each_vf_vport(esw, i, vport, esw->esw_funcs.num_vfs)
 		vport->info.link_state = MLX5_VPORT_ADMIN_STATE_DOWN;
+	if (mlx5_core_ec_sriov_enabled(esw->dev))
+		mlx5_esw_for_each_ec_vf_vport(esw, i, vport, esw->esw_funcs.num_ec_vfs)
+			vport->info.link_state = MLX5_VPORT_ADMIN_STATE_DOWN;
 
 	/* Uplink vport rep must load first. */
 	err = esw_offloads_load_rep(esw, MLX5_VPORT_UPLINK);
···
 			goto revert_inline_mode;
 		}
 	}
+	if (mlx5_core_ec_sriov_enabled(esw->dev)) {
+		mlx5_esw_for_each_ec_vf_vport(esw, i, vport, esw->esw_funcs.num_ec_vfs) {
+			err = mlx5_modify_nic_vport_min_inline(dev, vport->vport, mlx5_mode);
+			if (err) {
+				err_vport_num = vport->vport;
+				NL_SET_ERR_MSG_MOD(extack,
+						   "Failed to set min inline on vport");
+				goto revert_ec_vf_inline_mode;
+			}
+		}
+	}
 	return 0;
 
+revert_ec_vf_inline_mode:
+	mlx5_esw_for_each_ec_vf_vport(esw, i, vport, esw->esw_funcs.num_ec_vfs) {
+		if (vport->vport == err_vport_num)
+			break;
+		mlx5_modify_nic_vport_min_inline(dev,
+						 vport->vport,
+						 esw->offloads.inline_mode);
+	}
 revert_inline_mode:
 	mlx5_esw_for_each_host_func_vport(esw, i, vport, esw->esw_funcs.num_vfs) {
 		if (vport->vport == err_vport_num)

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

···
 #define MLX5_FW_REPORTER_VF_GRACEFUL_PERIOD 30000
 #define MLX5_FW_REPORTER_DEFAULT_GRACEFUL_PERIOD MLX5_FW_REPORTER_VF_GRACEFUL_PERIOD
 
-static void mlx5_fw_reporters_create(struct mlx5_core_dev *dev)
+void mlx5_fw_reporters_create(struct mlx5_core_dev *dev)
 {
 	struct mlx5_core_health *health = &dev->priv.health;
 	struct devlink *devlink = priv_to_devlink(dev);
···
 	}
 
 	health->fw_reporter =
-		devlink_health_reporter_create(devlink, &mlx5_fw_reporter_ops,
-					       0, dev);
+		devl_health_reporter_create(devlink, &mlx5_fw_reporter_ops,
+					    0, dev);
 	if (IS_ERR(health->fw_reporter))
 		mlx5_core_warn(dev, "Failed to create fw reporter, err = %ld\n",
 			       PTR_ERR(health->fw_reporter));
 
 	health->fw_fatal_reporter =
-		devlink_health_reporter_create(devlink,
-					       &mlx5_fw_fatal_reporter_ops,
-					       grace_period,
-					       dev);
+		devl_health_reporter_create(devlink,
+					    &mlx5_fw_fatal_reporter_ops,
+					    grace_period,
+					    dev);
 	if (IS_ERR(health->fw_fatal_reporter))
 		mlx5_core_warn(dev, "Failed to create fw fatal reporter, err = %ld\n",
 			       PTR_ERR(health->fw_fatal_reporter));
···
 {
 	struct mlx5_core_health *health = &dev->priv.health;
 
-	queue_work(health->wq, &health->fatal_report_work);
+	if (!mlx5_dev_is_lightweight(dev))
+		queue_work(health->wq, &health->fatal_report_work);
 }
 
 #define MLX5_MSEC_PER_HOUR (MSEC_PER_SEC * 60 * 60)
···
 
 int mlx5_health_init(struct mlx5_core_dev *dev)
 {
+	struct devlink *devlink = priv_to_devlink(dev);
 	struct mlx5_core_health *health;
 	char *name;
 
-	mlx5_fw_reporters_create(dev);
+	if (!mlx5_dev_is_lightweight(dev)) {
+		devl_lock(devlink);
+		mlx5_fw_reporters_create(dev);
+		devl_unlock(devlink);
+	}
 	mlx5_reporter_vnic_create(dev);
 
 	health = &dev->priv.health;

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

···
 	mlx5_devcom_unregister_device(dev->priv.devcom);
 }
 
-static int mlx5_function_setup(struct mlx5_core_dev *dev, bool boot, u64 timeout)
+static int mlx5_function_enable(struct mlx5_core_dev *dev, bool boot, u64 timeout)
 {
 	int err;
 
···
 		goto reclaim_boot_pages;
 	}
 
-	err = set_hca_ctrl(dev);
-	if (err) {
-		mlx5_core_err(dev, "set_hca_ctrl failed\n");
-		goto reclaim_boot_pages;
-	}
-
-	err = set_hca_cap(dev);
-	if (err) {
-		mlx5_core_err(dev, "set_hca_cap failed\n");
-		goto reclaim_boot_pages;
-	}
-
-	err = mlx5_satisfy_startup_pages(dev, 0);
-	if (err) {
-		mlx5_core_err(dev, "failed to allocate init pages\n");
-		goto reclaim_boot_pages;
-	}
-
-	err = mlx5_cmd_init_hca(dev, sw_owner_id);
-	if (err) {
-		mlx5_core_err(dev, "init hca failed\n");
-		goto reclaim_boot_pages;
-	}
-
-	mlx5_set_driver_version(dev);
-
-	err = mlx5_query_hca_caps(dev);
-	if (err) {
-		mlx5_core_err(dev, "query hca failed\n");
-		goto reclaim_boot_pages;
-	}
-	mlx5_start_health_fw_log_up(dev);
-
 	return 0;
 
 reclaim_boot_pages:
···
 	return err;
 }
 
-static int mlx5_function_teardown(struct mlx5_core_dev *dev, bool boot)
+static void mlx5_function_disable(struct mlx5_core_dev *dev, bool boot)
+{
+	mlx5_reclaim_startup_pages(dev);
+	mlx5_core_disable_hca(dev, 0);
+	mlx5_stop_health_poll(dev, boot);
+	mlx5_cmd_set_state(dev, MLX5_CMDIF_STATE_DOWN);
+	mlx5_cmd_cleanup(dev);
+}
+
+static int mlx5_function_open(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	err = set_hca_ctrl(dev);
+	if (err) {
+		mlx5_core_err(dev, "set_hca_ctrl failed\n");
+		return err;
+	}
+
+	err = set_hca_cap(dev);
+	if (err) {
+		mlx5_core_err(dev, "set_hca_cap failed\n");
+		return err;
+	}
+
+	err = mlx5_satisfy_startup_pages(dev, 0);
+	if (err) {
+		mlx5_core_err(dev, "failed to allocate init pages\n");
+		return err;
+	}
+
+	err = mlx5_cmd_init_hca(dev, sw_owner_id);
+	if (err) {
+		mlx5_core_err(dev, "init hca failed\n");
+		return err;
+	}
+
+	mlx5_set_driver_version(dev);
+
+	err = mlx5_query_hca_caps(dev);
+	if (err) {
+		mlx5_core_err(dev, "query hca failed\n");
+		return err;
+	}
+	mlx5_start_health_fw_log_up(dev);
+	return 0;
+}
+
+static int mlx5_function_close(struct mlx5_core_dev *dev)
 {
 	int err;
 
···
 		mlx5_core_err(dev, "tear_down_hca failed, skip cleanup\n");
 		return err;
 	}
-	mlx5_reclaim_startup_pages(dev);
-	mlx5_core_disable_hca(dev, 0);
-	mlx5_stop_health_poll(dev, boot);
-	mlx5_cmd_set_state(dev, MLX5_CMDIF_STATE_DOWN);
-	mlx5_cmd_cleanup(dev);
 
 	return 0;
+}
+
+static int mlx5_function_setup(struct mlx5_core_dev *dev, bool boot, u64 timeout)
+{
+	int err;
+
+	err = mlx5_function_enable(dev, boot, timeout);
+	if (err)
+		return err;
+
+	err = mlx5_function_open(dev);
+	if (err)
+		mlx5_function_disable(dev, boot);
+	return err;
+}
+
+static int mlx5_function_teardown(struct mlx5_core_dev *dev, bool boot)
+{
+	int err = mlx5_function_close(dev);
+
+	if (!err)
+		mlx5_function_disable(dev, boot);
+	return err;
 }
 
 static int mlx5_load(struct mlx5_core_dev *dev)
···
 	mlx5_put_uars_page(dev, dev->priv.uar);
 }
 
-int mlx5_init_one(struct mlx5_core_dev *dev)
+int mlx5_init_one_devl_locked(struct mlx5_core_dev *dev)
 {
-	struct devlink *devlink = priv_to_devlink(dev);
+	bool light_probe = mlx5_dev_is_lightweight(dev);
 	int err = 0;
 
-	devl_lock(devlink);
 	mutex_lock(&dev->intf_state_mutex);
 	dev->state = MLX5_DEVICE_STATE_UP;
 
···
 		goto function_teardown;
 	}
 
-	err = mlx5_devlink_params_register(priv_to_devlink(dev));
-	if (err)
-		goto err_devlink_params_reg;
+	/* In case of light_probe, mlx5_devlink is already registered.
+	 * Hence, don't register devlink again.
+	 */
+	if (!light_probe) {
+		err = mlx5_devlink_params_register(priv_to_devlink(dev));
+		if (err)
+			goto err_devlink_params_reg;
+	}
 
 	err = mlx5_load(dev);
 	if (err)
···
 		goto err_register;
 
 	mutex_unlock(&dev->intf_state_mutex);
-	devl_unlock(devlink);
 	return 0;
 
 err_register:
 	clear_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state);
 	mlx5_unload(dev);
 err_load:
-	mlx5_devlink_params_unregister(priv_to_devlink(dev));
+	if (!light_probe)
+		mlx5_devlink_params_unregister(priv_to_devlink(dev));
 err_devlink_params_reg:
 	mlx5_cleanup_once(dev);
 function_teardown:
···
 err_function:
 	dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
 	mutex_unlock(&dev->intf_state_mutex);
+	return err;
+}
+
+int mlx5_init_one(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	int err;
+
+	devl_lock(devlink);
+	err = mlx5_init_one_devl_locked(dev);
 	devl_unlock(devlink);
 	return err;
 }
···
 	devl_lock(devlink);
 	mlx5_unload_one_devl_locked(dev, suspend);
 	devl_unlock(devlink);
+}
+
+/* In case of light probe, we don't need a full query of hca_caps, but only the bellow caps.
+ * A full query of hca_caps will be done when the device will reload.
+ */
+static int mlx5_query_hca_caps_light(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL);
+	if (err)
+		return err;
+
+	if (MLX5_CAP_GEN(dev, eth_net_offloads)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ETHERNET_OFFLOADS);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, nic_flow_table) ||
+	    MLX5_CAP_GEN(dev, ipoib_enhanced_offloads)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_FLOW_TABLE);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN_64(dev, general_obj_types) &
+		MLX5_GENERAL_OBJ_TYPES_CAP_VIRTIO_NET_Q) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_VDPA_EMULATION);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+int mlx5_init_one_light(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	int err;
+
+	dev->state = MLX5_DEVICE_STATE_UP;
+	err = mlx5_function_enable(dev, true, mlx5_tout_ms(dev, FW_PRE_INIT_TIMEOUT));
+	if (err) {
+		mlx5_core_warn(dev, "mlx5_function_enable err=%d\n", err);
+		goto out;
+	}
+
+	err = mlx5_query_hca_caps_light(dev);
+	if (err) {
+		mlx5_core_warn(dev, "mlx5_query_hca_caps_light err=%d\n", err);
+		goto query_hca_caps_err;
+	}
+
+	devl_lock(devlink);
+	err = mlx5_devlink_params_register(priv_to_devlink(dev));
+	devl_unlock(devlink);
+	if (err) {
+		mlx5_core_warn(dev, "mlx5_devlink_param_reg err = %d\n", err);
+		goto query_hca_caps_err;
+	}
+
+	return 0;
+
+query_hca_caps_err:
+	mlx5_function_disable(dev, true);
+out:
+	dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
+	return err;
+}
+
+void mlx5_uninit_one_light(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+
+	devl_lock(devlink);
+	mlx5_devlink_params_unregister(priv_to_devlink(dev));
+	devl_unlock(devlink);
+	if (dev->state != MLX5_DEVICE_STATE_UP)
+		return;
+	mlx5_function_disable(dev, true);
+}
+
+/* xxx_light() function are used in order to configure the device without full
+ * init (light init). e.g.: There isn't a point in reload a device to light state.
+ * Hence, mlx5_load_one_light() isn't needed.
+ */
+
+void mlx5_unload_one_light(struct mlx5_core_dev *dev)
+{
+	if (dev->state != MLX5_DEVICE_STATE_UP)
+		return;
+	mlx5_function_disable(dev, false);
+	dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
 }
 
 static const int types[] = {
···
 	mlx5_drain_fw_reset(dev);
 	mlx5_drain_health_wq(dev);
 	devlink_unregister(devlink);
-	mlx5_sriov_disable(pdev);
+	mlx5_sriov_disable(pdev, false);
 	mlx5_thermal_uninit(dev);
 	mlx5_crdump_disable(dev);
 	mlx5_uninit_one(dev);

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

···
 int mlx5_sriov_attach(struct mlx5_core_dev *dev);
 void mlx5_sriov_detach(struct mlx5_core_dev *dev);
 int mlx5_core_sriov_configure(struct pci_dev *dev, int num_vfs);
-void mlx5_sriov_disable(struct pci_dev *pdev);
+void mlx5_sriov_disable(struct pci_dev *pdev, bool num_vf_change);
 int mlx5_core_sriov_set_msix_vec_count(struct pci_dev *vf, int msix_vec_count);
 int mlx5_core_enable_hca(struct mlx5_core_dev *dev, u16 func_id);
 int mlx5_core_disable_hca(struct mlx5_core_dev *dev, u16 func_id);
···
 void mlx5_detach_device(struct mlx5_core_dev *dev, bool suspend);
 int mlx5_register_device(struct mlx5_core_dev *dev);
 void mlx5_unregister_device(struct mlx5_core_dev *dev);
+void mlx5_dev_set_lightweight(struct mlx5_core_dev *dev);
+bool mlx5_dev_is_lightweight(struct mlx5_core_dev *dev);
 struct mlx5_core_dev *mlx5_get_next_phys_dev_lag(struct mlx5_core_dev *dev);
 void mlx5_dev_list_lock(void);
 void mlx5_dev_list_unlock(void);
 int mlx5_dev_list_trylock(void);
 
+void mlx5_fw_reporters_create(struct mlx5_core_dev *dev);
 int mlx5_query_mtpps(struct mlx5_core_dev *dev, u32 *mtpps, u32 mtpps_size);
 int mlx5_set_mtpps(struct mlx5_core_dev *mdev, u32 *mtpps, u32 mtpps_size);
 int mlx5_query_mtppse(struct mlx5_core_dev *mdev, u8 pin, u8 *arm, u8 *mode);
···
 int mlx5_mdev_init(struct mlx5_core_dev *dev, int profile_idx);
 void mlx5_mdev_uninit(struct mlx5_core_dev *dev);
 int mlx5_init_one(struct mlx5_core_dev *dev);
+int mlx5_init_one_devl_locked(struct mlx5_core_dev *dev);
 void mlx5_uninit_one(struct mlx5_core_dev *dev);
 void mlx5_unload_one(struct mlx5_core_dev *dev, bool suspend);
 void mlx5_unload_one_devl_locked(struct mlx5_core_dev *dev, bool suspend);
 int mlx5_load_one(struct mlx5_core_dev *dev, bool recovery);
 int mlx5_load_one_devl_locked(struct mlx5_core_dev *dev, bool recovery);
+int mlx5_init_one_light(struct mlx5_core_dev *dev);
+void mlx5_uninit_one_light(struct mlx5_core_dev *dev);
+void mlx5_unload_one_light(struct mlx5_core_dev *dev);
 
-int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap, u16 function_id,
+int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap, u16 vport,
 				  u16 opmod);
-#define mlx5_vport_get_other_func_general_cap(dev, fid, out)		\
-	mlx5_vport_get_other_func_cap(dev, fid, out, MLX5_CAP_GENERAL)
+#define mlx5_vport_get_other_func_general_cap(dev, vport, out)		\
+	mlx5_vport_get_other_func_cap(dev, vport, out, MLX5_CAP_GENERAL)
 
 void mlx5_events_work_enqueue(struct mlx5_core_dev *dev, struct work_struct *work);
 static inline u32 mlx5_sriov_get_vf_total_msix(struct pci_dev *pdev)
···
 bool mlx5_vnet_supported(struct mlx5_core_dev *dev);
 bool mlx5_same_hw_devs(struct mlx5_core_dev *dev, struct mlx5_core_dev *peer_dev);
 
+static inline u16 mlx5_core_ec_vf_vport_base(const struct mlx5_core_dev *dev)
+{
+	return MLX5_CAP_GEN_2(dev, ec_vf_vport_base);
+}
+
+static inline u16 mlx5_core_ec_sriov_enabled(const struct mlx5_core_dev *dev)
+{
+	return mlx5_core_is_ecpf(dev) && mlx5_core_ec_vf_vport_base(dev);
+}
+
+static inline bool mlx5_core_is_ec_vf_vport(const struct mlx5_core_dev *dev, u16 vport_num)
+{
+	int base_vport = mlx5_core_ec_vf_vport_base(dev);
+	int max_vport = base_vport + mlx5_core_max_ec_vfs(dev);
+
+	if (!mlx5_core_ec_sriov_enabled(dev))
+		return false;
+
+	return (vport_num >= base_vport && vport_num < max_vport);
+}
 #endif /* __MLX5_CORE_H__ */

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

···
 	if (!func_id)
 		return mlx5_core_is_ecpf(dev) && !ec_function ? MLX5_HOST_PF : MLX5_PF;
 
-	return func_id <= mlx5_core_max_vfs(dev) ?  MLX5_VF : MLX5_SF;
+	if (func_id <= max(mlx5_core_max_vfs(dev), mlx5_core_max_ec_vfs(dev))) {
+		if (ec_function)
+			return MLX5_EC_VF;
+		else
+			return MLX5_VF;
+	}
+	return MLX5_SF;
 }
 
 static u32 mlx5_get_ec_function(u32 function)
···
 	WARN(dev->priv.page_counters[MLX5_HOST_PF],
 	     "External host PF FW pages counter is %d after reclaiming all pages\n",
 	     dev->priv.page_counters[MLX5_HOST_PF]);
+	WARN(dev->priv.page_counters[MLX5_EC_VF],
+	     "EC VFs FW pages counter is %d after reclaiming all pages\n",
+	     dev->priv.page_counters[MLX5_EC_VF]);
 
 	return 0;
 }

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

···
 	struct mlx5_irq_pool *sf_comp_pool;
 };
 
+static int mlx5_core_func_to_vport(const struct mlx5_core_dev *dev,
+				   int func,
+				   bool ec_vf_func)
+{
+	if (!ec_vf_func)
+		return func;
+	return mlx5_core_ec_vf_vport_base(dev) + func - 1;
+}
+
 /**
  * mlx5_get_default_msix_vec_count - Get the default number of MSI-X vectors
  *                                   to be ssigned to each VF.
···
 	int set_sz = MLX5_ST_SZ_BYTES(set_hca_cap_in);
 	void *hca_cap = NULL, *query_cap = NULL, *cap;
 	int num_vf_msix, min_msix, max_msix;
+	bool ec_vf_function;
+	int vport;
 	int ret;
 
 	num_vf_msix = MLX5_CAP_GEN_MAX(dev, num_total_dynamic_vf_msix);
···
 		goto out;
 	}
 
-	ret = mlx5_vport_get_other_func_general_cap(dev, function_id, query_cap);
+	ec_vf_function = mlx5_core_ec_sriov_enabled(dev);
+	vport = mlx5_core_func_to_vport(dev, function_id, ec_vf_function);
+	ret = mlx5_vport_get_other_func_general_cap(dev, vport, query_cap);
 	if (ret)
 		goto out;
 
···
 
 	MLX5_SET(set_hca_cap_in, hca_cap, opcode, MLX5_CMD_OP_SET_HCA_CAP);
 	MLX5_SET(set_hca_cap_in, hca_cap, other_function, 1);
+	MLX5_SET(set_hca_cap_in, hca_cap, ec_vf_function, ec_vf_function);
 	MLX5_SET(set_hca_cap_in, hca_cap, function_id, function_id);
 
 	MLX5_SET(set_hca_cap_in, hca_cap, op_mod,

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

···
 
 #include <linux/mlx5/driver.h>
 #include <linux/mlx5/device.h>
+#include <linux/mlx5/eswitch.h>
 #include "mlx5_core.h"
 #include "dev.h"
 #include "devlink.h"
···
 	mdev->priv.adev_idx = adev->id;
 	sf_dev->mdev = mdev;
 
+	/* Only local SFs do light probe */
+	if (MLX5_ESWITCH_MANAGER(sf_dev->parent_mdev))
+		mlx5_dev_set_lightweight(mdev);
+
 	err = mlx5_mdev_init(mdev, MLX5_SF_PROF);
 	if (err) {
 		mlx5_core_warn(mdev, "mlx5_mdev_init on err=%d\n", err);
···
 		goto remap_err;
 	}
 
-	err = mlx5_init_one(mdev);
+	if (MLX5_ESWITCH_MANAGER(sf_dev->parent_mdev))
+		err = mlx5_init_one_light(mdev);
+	else
+		err = mlx5_init_one(mdev);
 	if (err) {
 		mlx5_core_warn(mdev, "mlx5_init_one err=%d\n", err);
 		goto init_one_err;
···
 
 	mlx5_drain_health_wq(sf_dev->mdev);
 	devlink_unregister(devlink);
-	mlx5_uninit_one(sf_dev->mdev);
+	if (mlx5_dev_is_lightweight(sf_dev->mdev))
+		mlx5_uninit_one_light(sf_dev->mdev);
+	else
+		mlx5_uninit_one(sf_dev->mdev);
 	iounmap(sf_dev->mdev->iseg);
 	mlx5_mdev_uninit(sf_dev->mdev);
 	mlx5_devlink_free(devlink);

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

···
 #include "mlx5_irq.h"
 #include "eswitch.h"
 
-static int sriov_restore_guids(struct mlx5_core_dev *dev, int vf)
+static int sriov_restore_guids(struct mlx5_core_dev *dev, int vf, u16 func_id)
 {
 	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
 	struct mlx5_hca_vport_context *in;
···
 			!!(in->node_guid) * MLX5_HCA_VPORT_SEL_NODE_GUID |
 			!!(in->policy) * MLX5_HCA_VPORT_SEL_STATE_POLICY;
 
-		err = mlx5_core_modify_hca_vport_context(dev, 1, 1, vf + 1, in);
+		err = mlx5_core_modify_hca_vport_context(dev, 1, 1, func_id, in);
 		if (err)
 			mlx5_core_warn(dev, "modify vport context failed, unable to restore VF %d settings\n", vf);
 
···
 {
 	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
 	int err, vf, num_msix_count;
+	int vport_num;
 
 	err = mlx5_eswitch_enable(dev->priv.eswitch, num_vfs);
 	if (err) {
···
 
 		sriov->vfs_ctx[vf].enabled = 1;
 		if (MLX5_CAP_GEN(dev, port_type) == MLX5_CAP_PORT_TYPE_IB) {
-			err = sriov_restore_guids(dev, vf);
+			vport_num = mlx5_core_ec_sriov_enabled(dev) ?
+					mlx5_core_ec_vf_vport_base(dev) + vf
+					: vf + 1;
+			err = sriov_restore_guids(dev, vf, vport_num);
 			if (err) {
 				mlx5_core_warn(dev,
 					       "failed to restore VF %d settings, err %d\n",
···
 }
 
 static void
-mlx5_device_disable_sriov(struct mlx5_core_dev *dev, int num_vfs, bool clear_vf)
+mlx5_device_disable_sriov(struct mlx5_core_dev *dev, int num_vfs, bool clear_vf, bool num_vf_change)
 {
 	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
+	bool wait_for_ec_vf_pages = true;
+	bool wait_for_vf_pages = true;
 	int err;
 	int vf;
 
···
 
 	mlx5_eswitch_disable_sriov(dev->priv.eswitch, clear_vf);
 
+	/* There are a number of scenarios when SRIOV is being disabled:
+	 *     1. VFs or ECVFs had been created, and now set back to 0 (num_vf_change == true).
+	 *		- If EC SRIOV is enabled then this flow is happening on the
+	 *		  embedded platform, wait for only EC VF pages.
+	 *		- If EC SRIOV is not enabled this flow is happening on non-embedded
+	 *		  platform, wait for the VF pages.
+	 *
+	 *     2. The driver is being unloaded. In this case wait for all pages.
+	 */
+	if (num_vf_change) {
+		if (mlx5_core_ec_sriov_enabled(dev))
+			wait_for_vf_pages = false;
+		else
+			wait_for_ec_vf_pages = false;
+	}
+
+	if (wait_for_ec_vf_pages && mlx5_wait_for_pages(dev, &dev->priv.page_counters[MLX5_EC_VF]))
+		mlx5_core_warn(dev, "timeout reclaiming EC VFs pages\n");
+
 	/* For ECPFs, skip waiting for host VF pages until ECPF is destroyed */
 	if (mlx5_core_is_ecpf(dev))
 		return;
 
-	if (mlx5_wait_for_pages(dev, &dev->priv.page_counters[MLX5_VF]))
+	if (wait_for_vf_pages && mlx5_wait_for_pages(dev, &dev->priv.page_counters[MLX5_VF]))
 		mlx5_core_warn(dev, "timeout reclaiming VFs pages\n");
 }
 
···
 	err = pci_enable_sriov(pdev, num_vfs);
 	if (err) {
 		mlx5_core_warn(dev, "pci_enable_sriov failed : %d\n", err);
-		mlx5_device_disable_sriov(dev, num_vfs, true);
+		mlx5_device_disable_sriov(dev, num_vfs, true, true);
 	}
 	return err;
 }
 
-void mlx5_sriov_disable(struct pci_dev *pdev)
+void mlx5_sriov_disable(struct pci_dev *pdev, bool num_vf_change)
 {
 	struct mlx5_core_dev *dev  = pci_get_drvdata(pdev);
 	struct devlink *devlink = priv_to_devlink(dev);
···
 
 	pci_disable_sriov(pdev);
 	devl_lock(devlink);
-	mlx5_device_disable_sriov(dev, num_vfs, true);
+	mlx5_device_disable_sriov(dev, num_vfs, true, num_vf_change);
 	devl_unlock(devlink);
 }
 
···
 	if (num_vfs)
 		err = mlx5_sriov_enable(pdev, num_vfs);
 	else
-		mlx5_sriov_disable(pdev);
+		mlx5_sriov_disable(pdev, true);
 
 	if (!err)
 		sriov->num_vfs = num_vfs;
···
 	if (!mlx5_core_is_pf(dev))
 		return;
 
-	mlx5_device_disable_sriov(dev, pci_num_vf(dev->pdev), false);
+	mlx5_device_disable_sriov(dev, pci_num_vf(dev->pdev), false, false);
 }
 
 static u16 mlx5_get_max_vfs(struct mlx5_core_dev *dev)
···
 	total_vfs = pci_sriov_get_totalvfs(pdev);
 	sriov->max_vfs = mlx5_get_max_vfs(dev);
 	sriov->num_vfs = pci_num_vf(pdev);
+	sriov->max_ec_vfs = mlx5_core_ec_sriov_enabled(dev) ? pci_sriov_get_totalvfs(dev->pdev) : 0;
 	sriov->vfs_ctx = kcalloc(total_vfs, sizeof(*sriov->vfs_ctx), GFP_KERNEL);
 	if (!sriov->vfs_ctx)
 		return -ENOMEM;

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

···
 }
 EXPORT_SYMBOL_GPL(mlx5_query_nic_system_image_guid);
 
-int mlx5_vport_get_other_func_cap(struct mlx5_core_dev *dev, u16 function_id, void *out,
+static int mlx5_vport_to_func_id(const struct mlx5_core_dev *dev, u16 vport, bool ec_vf_func)
+{
+	return ec_vf_func ? vport - mlx5_core_ec_vf_vport_base(dev)
+			  : vport;
+}
+
+int mlx5_vport_get_other_func_cap(struct mlx5_core_dev *dev, u16 vport, void *out,
 				  u16 opmod)
 {
+	bool ec_vf_func = mlx5_core_is_ec_vf_vport(dev, vport);
 	u8 in[MLX5_ST_SZ_BYTES(query_hca_cap_in)] = {};
 
 	opmod = (opmod << 1) | (HCA_CAP_OPMOD_GET_MAX & 0x01);
 	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
 	MLX5_SET(query_hca_cap_in, in, op_mod, opmod);
-	MLX5_SET(query_hca_cap_in, in, function_id, function_id);
+	MLX5_SET(query_hca_cap_in, in, function_id, mlx5_vport_to_func_id(dev, vport, ec_vf_func));
 	MLX5_SET(query_hca_cap_in, in, other_function, true);
+	MLX5_SET(query_hca_cap_in, in, ec_vf_function, ec_vf_func);
 	return mlx5_cmd_exec_inout(dev, query_hca_cap, in, out);
 }
 EXPORT_SYMBOL_GPL(mlx5_vport_get_other_func_cap);
 
 int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap,
-				  u16 function_id, u16 opmod)
+				  u16 vport, u16 opmod)
 {
+	bool ec_vf_func = mlx5_core_is_ec_vf_vport(dev, vport);
 	int set_sz = MLX5_ST_SZ_BYTES(set_hca_cap_in);
 	void *set_hca_cap;
 	void *set_ctx;
···
 	MLX5_SET(set_hca_cap_in, set_ctx, op_mod, opmod << 1);
 	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx, capability);
 	memcpy(set_hca_cap, hca_cap, MLX5_ST_SZ_BYTES(cmd_hca_cap));
-	MLX5_SET(set_hca_cap_in, set_ctx, function_id, function_id);
+	MLX5_SET(set_hca_cap_in, set_ctx, function_id,
+		 mlx5_vport_to_func_id(dev, vport, ec_vf_func));
 	MLX5_SET(set_hca_cap_in, set_ctx, other_function, true);
+	MLX5_SET(set_hca_cap_in, set_ctx, ec_vf_function, ec_vf_func);
 	ret = mlx5_cmd_exec_in(dev, set_hca_cap, set_ctx);
 
 	kfree(set_ctx);

include/linux/mlx5/driver.h

···
 	struct mlx5_vf_context	*vfs_ctx;
 	int			num_vfs;
 	u16			max_vfs;
+	u16			max_ec_vfs;
 };
 
 struct mlx5_fc_pool {
···
 	MLX5_VF,
 	MLX5_SF,
 	MLX5_HOST_PF,
+	MLX5_EC_VF,
 	MLX5_FUNC_TYPE_NUM,
 };
 
···
 static inline u16 mlx5_core_max_vfs(const struct mlx5_core_dev *dev)
 {
 	return dev->priv.sriov.max_vfs;
+}
+
+static inline u16 mlx5_core_max_ec_vfs(const struct mlx5_core_dev *dev)
+{
+	return dev->priv.sriov.max_ec_vfs;
 }
 
 static inline int mlx5_get_gid_table_len(u16 param)

include/linux/mlx5/mlx5_ifc.h

···
 	u8	   ts_cqe_metadata_size2wqe_counter[0x5];
 	u8	   reserved_at_250[0x10];
 
-	u8	   reserved_at_260[0x5a0];
+	u8	   reserved_at_260[0x120];
+	u8	   reserved_at_380[0x10];
+	u8	   ec_vf_vport_base[0x10];
+	u8	   reserved_at_3a0[0x460];
 };
 
 enum mlx5_ifc_flow_destination_type {
···
 	u8         op_mod[0x10];
 
 	u8         other_function[0x1];
-	u8         reserved_at_41[0xf];
+	u8         ec_vf_function[0x1];
+	u8         reserved_at_42[0xe];
 	u8         function_id[0x10];
 
 	u8         reserved_at_60[0x20];
···
 	u8         op_mod[0x10];
 
 	u8         other_function[0x1];
-	u8         reserved_at_41[0xf];
+	u8         ec_vf_function[0x1];
+	u8         reserved_at_42[0xe];
 	u8         function_id[0x10];
 
 	u8         reserved_at_60[0x20];

include/linux/mlx5/vport.h

···
 int mlx5_nic_vport_unaffiliate_multiport(struct mlx5_core_dev *port_mdev);
 
 u64 mlx5_query_nic_system_image_guid(struct mlx5_core_dev *mdev);
-int mlx5_vport_get_other_func_cap(struct mlx5_core_dev *dev, u16 function_id, void *out,
+int mlx5_vport_get_other_func_cap(struct mlx5_core_dev *dev, u16 vport, void *out,
 				  u16 opmod);
 #endif /* __MLX5_VPORT_H__ */