[GIT PULL][mlx5-next 0/4] Cached vhca id and adjacent function vports

[GIT PULL][mlx5-next 0/4] Cached vhca id and adjacent function vports

[Saeed Mahameed]

From: Saeed Mahameed <saeedm@nvidia.com>

Hi Jakub, Jason,

This pull request introduces a preparation patchset for caching vhca_id 
and needed HW bits for the upcoming netdev/eswitch series to support
adjacent function vports.

For more information please see tag log below.

The following changes since commit 8f5ae30d69d7543eee0d70083daf4de8fe15d585:

  Linux 6.17-rc1 (2025-08-10 19:41:16 +0300)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux.git tags/mlx5-next-vhca-id

for you to fetch changes up to 40653f280b2640e5caa94eeedee43e0f1df97704:

  {rdma,net}/mlx5: export mlx5_vport_get_vhca_id (2025-08-15 12:17:47 -0700)

----------------------------------------------------------------
mlx5-next-vhca-id

A preparation patchset for adjacent function vports.

Adjacent functions can delegate their SR-IOV VFs to sibling PFs,
allowing for more flexible and scalable management in multi-host and
ECPF-to-host scenarios. Adjacent vports can be managed by the management
PF via their unique vhca id and can't be managed by function index as the
index can conflict with the local vports/vfs.

This series provides:

- Use the cached vcha id instead of querying it every time from fw
- Query hca cap using vhca id instead of function id when FW supports it
- Add HW capabilities and required definitions for adjacent function vports

----------------------------------------------------------------
Saeed Mahameed (4):
      net/mlx5: mlx5_ifc, Add hardware definitions needed for adjacent vports
      net/mlx5: E-Switch, Cache vport vhca id on first cap query
      net/mlx5: E-Switch, Set/Query hca cap via vhca id
      {rdma,net}/mlx5: export mlx5_vport_get_vhca_id

 drivers/infiniband/hw/mlx5/std_types.c                        |  27 ++++-----------------------
 drivers/net/ethernet/mellanox/mlx5/core/diag/reporter_vnic.c  |   2 ++
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.c             |  18 ++++++++++++++++--
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h             |  20 ++++++++++++++++++--
 drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c    |  34 +++++++++++++++++-----------------
 drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h           |   2 --
 drivers/net/ethernet/mellanox/mlx5/core/steering/hws/cmd.c    |  16 +++++++++++-----
 drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_cmd.c |  16 ++++++++++++----
 drivers/net/ethernet/mellanox/mlx5/core/vport.c               |  58 ++++++++++++++++++++++++++++++++++++++++++++++++++++------
 include/linux/mlx5/mlx5_ifc.h                                 | 133 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++----
 include/linux/mlx5/vport.h                                    |   2 ++
 11 files changed, 263 insertions(+), 65 deletions(-)

-- 
2.50.1

[PATCH mlx5-next 1/4] net/mlx5: mlx5_ifc, Add hardware definitions needed for adjacent vports

[Saeed Mahameed]

From: Saeed Mahameed <saeedm@nvidia.com>

Next patches will implement the discovery and creation of adjacent
functions vports, this patch introduces the hardware structures
definitions needed for the driver implementation.

Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Reviewed-by: Jack Morgenstein <jackm@nvidia.com>
Signed-off-by: Alexei Lazar <alazar@nvidia.com>
---
 include/linux/mlx5/mlx5_ifc.h | 133 +++++++++++++++++++++++++++++++++-
 1 file changed, 129 insertions(+), 4 deletions(-)

diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index 8360d9011d4f..44d497272162 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -189,6 +189,9 @@ enum {
 	MLX5_CMD_OP_QUERY_XRQ_ERROR_PARAMS        = 0x727,
 	MLX5_CMD_OP_RELEASE_XRQ_ERROR             = 0x729,
 	MLX5_CMD_OP_MODIFY_XRQ                    = 0x72a,
+	MLX5_CMD_OPCODE_QUERY_DELEGATED_VHCA      = 0x732,
+	MLX5_CMD_OPCODE_CREATE_ESW_VPORT          = 0x733,
+	MLX5_CMD_OPCODE_DESTROY_ESW_VPORT         = 0x734,
 	MLX5_CMD_OP_QUERY_ESW_FUNCTIONS           = 0x740,
 	MLX5_CMD_OP_QUERY_VPORT_STATE             = 0x750,
 	MLX5_CMD_OP_MODIFY_VPORT_STATE            = 0x751,
@@ -2207,7 +2210,19 @@ struct mlx5_ifc_cmd_hca_cap_2_bits {
 
 	u8	   reserved_at_440[0x8];
 	u8	   max_num_eqs_24b[0x18];
-	u8	   reserved_at_460[0x3a0];
+
+	u8         reserved_at_460[0x160];
+
+	u8         query_adjacent_functions_id[0x1];
+	u8         ingress_egress_esw_vport_connect[0x1];
+	u8         function_id_type_vhca_id[0x1];
+	u8         reserved_at_5c3[0xd];
+	u8         delegate_vhca_management_profiles[0x10];
+
+	u8         delegated_vhca_max[0x10];
+	u8         delegate_vhca_max[0x10];
+
+	u8         reserved_at_600[0x200];
 };
 
 enum mlx5_ifc_flow_destination_type {
@@ -5159,7 +5174,9 @@ struct mlx5_ifc_set_hca_cap_in_bits {
 
 	u8         other_function[0x1];
 	u8         ec_vf_function[0x1];
-	u8         reserved_at_42[0xe];
+	u8         reserved_at_42[0x1];
+	u8         function_id_type[0x1];
+	u8         reserved_at_44[0xc];
 	u8         function_id[0x10];
 
 	u8         reserved_at_60[0x20];
@@ -6357,7 +6374,9 @@ struct mlx5_ifc_query_hca_cap_in_bits {
 
 	u8         other_function[0x1];
 	u8         ec_vf_function[0x1];
-	u8         reserved_at_42[0xe];
+	u8         reserved_at_42[0x1];
+	u8         function_id_type[0x1];
+	u8         reserved_at_44[0xc];
 	u8         function_id[0x10];
 
 	u8         reserved_at_60[0x20];
@@ -6983,6 +7002,28 @@ struct mlx5_ifc_query_esw_vport_context_in_bits {
 	u8         reserved_at_60[0x20];
 };
 
+struct mlx5_ifc_destroy_esw_vport_out_bits {
+	u8         status[0x8];
+	u8         reserved_at_8[0x18];
+
+	u8         syndrome[0x20];
+
+	u8         reserved_at_40[0x20];
+};
+
+struct mlx5_ifc_destroy_esw_vport_in_bits {
+	u8         opcode[0x10];
+	u8         uid[0x10];
+
+	u8         reserved_at_20[0x10];
+	u8         op_mod[0x10];
+
+	u8         reserved_at_40[0x10];
+	u8         vport_num[0x10];
+
+	u8         reserved_at_60[0x20];
+};
+
 struct mlx5_ifc_modify_esw_vport_context_out_bits {
 	u8         status[0x8];
 	u8         reserved_at_8[0x18];
@@ -7484,6 +7525,85 @@ struct mlx5_ifc_query_adapter_in_bits {
 	u8         reserved_at_40[0x40];
 };
 
+struct mlx5_ifc_function_vhca_rid_info_reg_bits {
+	u8         host_number[0x8];
+	u8         host_pci_device_function[0x8];
+	u8         host_pci_bus[0x8];
+	u8         reserved_at_18[0x3];
+	u8         pci_bus_assigned[0x1];
+	u8         function_type[0x4];
+
+	u8         parent_pci_device_function[0x8];
+	u8         parent_pci_bus[0x8];
+	u8         vhca_id[0x10];
+
+	u8         reserved_at_40[0x10];
+	u8         function_id[0x10];
+
+	u8         reserved_at_60[0x20];
+};
+
+struct mlx5_ifc_delegated_function_vhca_rid_info_bits {
+	struct mlx5_ifc_function_vhca_rid_info_reg_bits function_vhca_rid_info;
+
+	u8         reserved_at_80[0x18];
+	u8         manage_profile[0x8];
+
+	u8         reserved_at_a0[0x60];
+};
+
+struct mlx5_ifc_query_delegated_vhca_out_bits {
+	u8         status[0x8];
+	u8         reserved_at_8[0x18];
+
+	u8         syndrome[0x20];
+
+	u8         reserved_at_40[0x20];
+
+	u8         reserved_at_60[0x10];
+	u8         functions_count[0x10];
+
+	u8         reserved_at_80[0x80];
+
+	struct mlx5_ifc_delegated_function_vhca_rid_info_bits
+			delegated_function_vhca_rid_info[];
+};
+
+struct mlx5_ifc_query_delegated_vhca_in_bits {
+	u8         opcode[0x10];
+	u8         uid[0x10];
+
+	u8         reserved_at_20[0x10];
+	u8         op_mod[0x10];
+
+	u8         reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_create_esw_vport_out_bits {
+	u8         status[0x8];
+	u8         reserved_at_8[0x18];
+
+	u8         syndrome[0x20];
+
+	u8         reserved_at_40[0x20];
+
+	u8         reserved_at_60[0x10];
+	u8         vport_num[0x10];
+};
+
+struct mlx5_ifc_create_esw_vport_in_bits {
+	u8         opcode[0x10];
+	u8         reserved_at_10[0x10];
+
+	u8         reserved_at_20[0x10];
+	u8         op_mod[0x10];
+
+	u8         reserved_at_40[0x10];
+	u8         managed_vhca_id[0x10];
+
+	u8         reserved_at_60[0x20];
+};
+
 struct mlx5_ifc_qp_2rst_out_bits {
 	u8         status[0x8];
 	u8         reserved_at_8[0x18];
@@ -7611,7 +7731,12 @@ struct mlx5_ifc_modify_vport_state_in_bits {
 	u8         reserved_at_41[0xf];
 	u8         vport_number[0x10];
 
-	u8         reserved_at_60[0x18];
+	u8         reserved_at_60[0x10];
+	u8         ingress_connect[0x1];
+	u8         egress_connect[0x1];
+	u8         ingress_connect_valid[0x1];
+	u8         egress_connect_valid[0x1];
+	u8         reserved_at_74[0x4];
 	u8         admin_state[0x4];
 	u8         reserved_at_7c[0x4];
 };
-- 
2.50.1

[PATCH mlx5-next 2/4] net/mlx5: E-Switch, Cache vport vhca id on first cap query

[Saeed Mahameed]

From: Saeed Mahameed <saeedm@nvidia.com>

We need vhca_id to set up the vhca_id to vport mapping for every vport,
for that we query the firmware in mlx5_esw_vport_vhca_id_set, and it is
redundant since in esw_vport_setup, we already query hca caps which has
the vhca_id, cache it there and save 2 extra fw queries per vport.

Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Alexei Lazar <alazar@nvidia.com>
Reviewed-by: Feng Liu <feliu@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/eswitch.c |  6 ++--
 .../net/ethernet/mellanox/mlx5/core/eswitch.h | 12 +++++--
 .../mellanox/mlx5/core/eswitch_offloads.c     | 34 +++++++++----------
 3 files changed, 31 insertions(+), 21 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index 4917d185d0c3..eeffe9c4aa56 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -820,6 +820,7 @@ static int mlx5_esw_vport_caps_get(struct mlx5_eswitch *esw, struct mlx5_vport *
 
 	hca_caps = MLX5_ADDR_OF(query_hca_cap_out, query_ctx, capability);
 	vport->info.roce_enabled = MLX5_GET(cmd_hca_cap, hca_caps, roce);
+	vport->vhca_id = MLX5_GET(cmd_hca_cap, hca_caps, vhca_id);
 
 	if (!MLX5_CAP_GEN_MAX(esw->dev, hca_cap_2))
 		goto out_free;
@@ -929,7 +930,7 @@ int mlx5_esw_vport_enable(struct mlx5_eswitch *esw, struct mlx5_vport *vport,
 
 	if (!mlx5_esw_is_manager_vport(esw, vport_num) &&
 	    MLX5_CAP_GEN(esw->dev, vhca_resource_manager)) {
-		ret = mlx5_esw_vport_vhca_id_set(esw, vport_num);
+		ret = mlx5_esw_vport_vhca_id_map(esw, vport);
 		if (ret)
 			goto err_vhca_mapping;
 	}
@@ -973,7 +974,7 @@ void mlx5_esw_vport_disable(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
 
 	if (!mlx5_esw_is_manager_vport(esw, vport_num) &&
 	    MLX5_CAP_GEN(esw->dev, vhca_resource_manager))
-		mlx5_esw_vport_vhca_id_clear(esw, vport_num);
+		mlx5_esw_vport_vhca_id_unmap(esw, vport);
 
 	if (vport->vport != MLX5_VPORT_PF &&
 	    (vport->info.ipsec_crypto_enabled || vport->info.ipsec_packet_enabled))
@@ -1710,6 +1711,7 @@ static int mlx5_esw_vport_alloc(struct mlx5_eswitch *esw,
 	vport->vport = vport_num;
 	vport->index = index;
 	vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO;
+	vport->vhca_id = MLX5_VHCA_ID_INVALID;
 	INIT_WORK(&vport->vport_change_handler, esw_vport_change_handler);
 	err = xa_insert(&esw->vports, vport_num, vport, GFP_KERNEL);
 	if (err)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index b0b8ef3ec3c4..7f6bfaae5f5f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -197,6 +197,11 @@ static inline struct mlx5_vport *mlx5_devlink_port_vport_get(struct devlink_port
 	return mlx5_devlink_port_get(dl_port)->vport;
 }
 
+#define MLX5_VHCA_ID_INVALID (-1)
+
+#define MLX5_VPORT_INVAL_VHCA_ID(vport) \
+	((vport)->vhca_id == MLX5_VHCA_ID_INVALID)
+
 struct mlx5_vport {
 	struct mlx5_core_dev    *dev;
 	struct hlist_head       uc_list[MLX5_L2_ADDR_HASH_SIZE];
@@ -209,6 +214,7 @@ struct mlx5_vport {
 	struct vport_egress     egress;
 	u32                     default_metadata;
 	u32                     metadata;
+	int                     vhca_id;
 
 	struct mlx5_vport_info  info;
 
@@ -822,8 +828,10 @@ struct devlink_port *mlx5_esw_offloads_devlink_port(struct mlx5_eswitch *esw, u1
 
 int mlx5_esw_sf_max_hpf_functions(struct mlx5_core_dev *dev, u16 *max_sfs, u16 *sf_base_id);
 
-int mlx5_esw_vport_vhca_id_set(struct mlx5_eswitch *esw, u16 vport_num);
-void mlx5_esw_vport_vhca_id_clear(struct mlx5_eswitch *esw, u16 vport_num);
+int mlx5_esw_vport_vhca_id_map(struct mlx5_eswitch *esw,
+			       struct mlx5_vport *vport);
+void mlx5_esw_vport_vhca_id_unmap(struct mlx5_eswitch *esw,
+				  struct mlx5_vport *vport);
 int mlx5_eswitch_vhca_id_to_vport(struct mlx5_eswitch *esw, u16 vhca_id, u16 *vport_num);
 
 /**
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index bee906661282..19decaa8a96e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -4161,23 +4161,28 @@ u32 mlx5_eswitch_get_vport_metadata_for_match(struct mlx5_eswitch *esw,
 }
 EXPORT_SYMBOL(mlx5_eswitch_get_vport_metadata_for_match);
 
-int mlx5_esw_vport_vhca_id_set(struct mlx5_eswitch *esw, u16 vport_num)
+int mlx5_esw_vport_vhca_id_map(struct mlx5_eswitch *esw,
+			       struct mlx5_vport *vport)
 {
 	u16 *old_entry, *vhca_map_entry, vhca_id;
-	int err;
 
-	err = mlx5_vport_get_vhca_id(esw->dev, vport_num, &vhca_id);
-	if (err) {
-		esw_warn(esw->dev, "Getting vhca_id for vport failed (vport=%u,err=%d)\n",
-			 vport_num, err);
-		return err;
+	if (WARN_ONCE(MLX5_VPORT_INVAL_VHCA_ID(vport),
+		      "vport %d vhca_id is not set", vport->vport)) {
+		int err;
+
+		err = mlx5_vport_get_vhca_id(vport->dev, vport->vport,
+					     &vhca_id);
+		if (err)
+			return err;
+		vport->vhca_id = vhca_id;
 	}
 
+	vhca_id = vport->vhca_id;
 	vhca_map_entry = kmalloc(sizeof(*vhca_map_entry), GFP_KERNEL);
 	if (!vhca_map_entry)
 		return -ENOMEM;
 
-	*vhca_map_entry = vport_num;
+	*vhca_map_entry = vport->vport;
 	old_entry = xa_store(&esw->offloads.vhca_map, vhca_id, vhca_map_entry, GFP_KERNEL);
 	if (xa_is_err(old_entry)) {
 		kfree(vhca_map_entry);
@@ -4187,17 +4192,12 @@ int mlx5_esw_vport_vhca_id_set(struct mlx5_eswitch *esw, u16 vport_num)
 	return 0;
 }
 
-void mlx5_esw_vport_vhca_id_clear(struct mlx5_eswitch *esw, u16 vport_num)
+void mlx5_esw_vport_vhca_id_unmap(struct mlx5_eswitch *esw,
+				  struct mlx5_vport *vport)
 {
-	u16 *vhca_map_entry, vhca_id;
-	int err;
-
-	err = mlx5_vport_get_vhca_id(esw->dev, vport_num, &vhca_id);
-	if (err)
-		esw_warn(esw->dev, "Getting vhca_id for vport failed (vport=%hu,err=%d)\n",
-			 vport_num, err);
+	u16 *vhca_map_entry;
 
-	vhca_map_entry = xa_erase(&esw->offloads.vhca_map, vhca_id);
+	vhca_map_entry = xa_erase(&esw->offloads.vhca_map, vport->vhca_id);
 	kfree(vhca_map_entry);
 }
 
-- 
2.50.1

[PATCH mlx5-next 3/4] net/mlx5: E-Switch, Set/Query hca cap via vhca id

[Saeed Mahameed]

From: Saeed Mahameed <saeedm@nvidia.com>

Dynamically created vports require vhca id as input to set/query other
vport hca cap, when FW is capable and the vhca id of a vport is valid
use it instead of the local function id.

Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Adithya Jayachandran <ajayachandra@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Reviewed-by: Feng Liu <feliu@nvidia.com>
Reviewed-by: William Tu <witu@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/eswitch.c | 12 +++++
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  8 +++
 .../net/ethernet/mellanox/mlx5/core/vport.c   | 53 +++++++++++++++++--
 3 files changed, 68 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index eeffe9c4aa56..21c42138d93c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -840,6 +840,18 @@ static int mlx5_esw_vport_caps_get(struct mlx5_eswitch *esw, struct mlx5_vport *
 	return err;
 }
 
+bool mlx5_esw_vport_vhca_id(struct mlx5_eswitch *esw, u16 vportn, u16 *vhca_id)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(esw, vportn);
+	if (IS_ERR(vport) || MLX5_VPORT_INVAL_VHCA_ID(vport))
+		return false;
+
+	*vhca_id = vport->vhca_id;
+	return true;
+}
+
 static int esw_vport_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
 {
 	bool vst_mode_steering = esw_vst_mode_is_steering(esw);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 7f6bfaae5f5f..f47389629c62 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -833,6 +833,7 @@ int mlx5_esw_vport_vhca_id_map(struct mlx5_eswitch *esw,
 void mlx5_esw_vport_vhca_id_unmap(struct mlx5_eswitch *esw,
 				  struct mlx5_vport *vport);
 int mlx5_eswitch_vhca_id_to_vport(struct mlx5_eswitch *esw, u16 vhca_id, u16 *vport_num);
+bool mlx5_esw_vport_vhca_id(struct mlx5_eswitch *esw, u16 vportn, u16 *vhca_id);
 
 /**
  * struct mlx5_esw_event_info - Indicates eswitch mode changed/changing.
@@ -973,6 +974,13 @@ static inline bool mlx5_eswitch_block_ipsec(struct mlx5_core_dev *dev)
 }
 
 static inline void mlx5_eswitch_unblock_ipsec(struct mlx5_core_dev *dev) {}
+
+static inline bool
+mlx5_esw_vport_vhca_id(struct mlx5_eswitch *esw, u16 vportn, u16 *vhca_id)
+{
+	return -EOPNOTSUPP;
+}
+
 #endif /* CONFIG_MLX5_ESWITCH */
 
 #endif /* __MLX5_ESWITCH_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/vport.c b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
index da5c24fc7b30..231bedc6a252 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/vport.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
@@ -36,6 +36,7 @@
 #include <linux/mlx5/vport.h>
 #include <linux/mlx5/eswitch.h>
 #include "mlx5_core.h"
+#include "eswitch.h"
 #include "sf/sf.h"
 
 /* Mutex to hold while enabling or disabling RoCE */
@@ -1189,18 +1190,44 @@ u64 mlx5_query_nic_system_image_guid(struct mlx5_core_dev *mdev)
 }
 EXPORT_SYMBOL_GPL(mlx5_query_nic_system_image_guid);
 
+static bool mlx5_vport_use_vhca_id_as_func_id(struct mlx5_core_dev *dev,
+					      u16 vport_num, u16 *vhca_id)
+{
+	if (!MLX5_CAP_GEN_2(dev, function_id_type_vhca_id))
+		return false;
+
+	return mlx5_esw_vport_vhca_id(dev->priv.eswitch, vport_num, vhca_id);
+}
+
 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)] = {};
+	u16 vhca_id = 0, function_id = 0;
+	bool ec_vf_func = false;
+
+	/* if this vport is referring to a vport on the ec PF (embedded cpu )
+	 * let the FW know which domain we are querying since vport numbers or
+	 * function_ids are not unique across the different PF domains,
+	 * unless we use vhca_id as the function_id below.
+	 */
+	ec_vf_func = mlx5_core_is_ec_vf_vport(dev, vport);
+	function_id = mlx5_vport_to_func_id(dev, vport, ec_vf_func);
+
+	if (mlx5_vport_use_vhca_id_as_func_id(dev, vport, &vhca_id)) {
+		MLX5_SET(query_hca_cap_in, in, function_id_type, 1);
+		function_id = vhca_id;
+		ec_vf_func = false;
+		mlx5_core_dbg(dev, "%s using vhca_id as function_id for vport %d vhca_id 0x%x\n",
+			      __func__, vport, vhca_id);
+	}
 
 	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, 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);
+	MLX5_SET(query_hca_cap_in, in, function_id, function_id);
 	return mlx5_cmd_exec_inout(dev, query_hca_cap, in, out);
 }
 EXPORT_SYMBOL_GPL(mlx5_vport_get_other_func_cap);
@@ -1233,8 +1260,9 @@ int mlx5_vport_get_vhca_id(struct mlx5_core_dev *dev, u16 vport, u16 *vhca_id)
 int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap,
 				  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);
+	u16 vhca_id = 0, function_id = 0;
+	bool ec_vf_func = false;
 	void *set_hca_cap;
 	void *set_ctx;
 	int ret;
@@ -1243,14 +1271,29 @@ int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap
 	if (!set_ctx)
 		return -ENOMEM;
 
+	/* if this vport is referring to a vport on the ec PF (embedded cpu )
+	 * let the FW know which domain we are querying since vport numbers or
+	 * function_ids are not unique across the different PF domains,
+	 * unless we use vhca_id as the function_id below.
+	 */
+	ec_vf_func = mlx5_core_is_ec_vf_vport(dev, vport);
+	function_id = mlx5_vport_to_func_id(dev, vport, ec_vf_func);
+
+	if (mlx5_vport_use_vhca_id_as_func_id(dev, vport, &vhca_id)) {
+		MLX5_SET(set_hca_cap_in, set_ctx, function_id_type, 1);
+		function_id = vhca_id;
+		ec_vf_func = false;
+		mlx5_core_dbg(dev, "%s using vhca_id as function_id for vport %d vhca_id 0x%x\n",
+			      __func__, vport, vhca_id);
+	}
+
 	MLX5_SET(set_hca_cap_in, set_ctx, opcode, MLX5_CMD_OP_SET_HCA_CAP);
 	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,
-		 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);
+	MLX5_SET(set_hca_cap_in, set_ctx, function_id, function_id);
 	ret = mlx5_cmd_exec_in(dev, set_hca_cap, set_ctx);
 
 	kfree(set_ctx);
-- 
2.50.1

[PATCH mlx5-next 4/4] {rdma,net}/mlx5: export mlx5_vport_get_vhca_id

[Saeed Mahameed]

From: Saeed Mahameed <saeedm@nvidia.com>

vhca id is already cached in the vport structure no need to query on
every mlx5 layer, use the mlx5_vport_get_vhca_id, where possible.

Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Alexei Lazar <alazar@nvidia.com>
Reviewed-by: Feng Liu <feliu@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/infiniband/hw/mlx5/std_types.c        | 27 +++----------------
 .../mellanox/mlx5/core/diag/reporter_vnic.c   |  2 ++
 .../ethernet/mellanox/mlx5/core/mlx5_core.h   |  2 --
 .../mellanox/mlx5/core/steering/hws/cmd.c     | 16 +++++++----
 .../mellanox/mlx5/core/steering/sws/dr_cmd.c  | 16 ++++++++---
 .../net/ethernet/mellanox/mlx5/core/vport.c   |  5 +++-
 include/linux/mlx5/vport.h                    |  2 ++
 7 files changed, 35 insertions(+), 35 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/std_types.c b/drivers/infiniband/hw/mlx5/std_types.c
index bdb568411091..2fcf553044e1 100644
--- a/drivers/infiniband/hw/mlx5/std_types.c
+++ b/drivers/infiniband/hw/mlx5/std_types.c
@@ -83,33 +83,14 @@ static int fill_vport_icm_addr(struct mlx5_core_dev *mdev, u16 vport,
 static int fill_vport_vhca_id(struct mlx5_core_dev *mdev, u16 vport,
 			      struct mlx5_ib_uapi_query_port *info)
 {
-	size_t out_sz = MLX5_ST_SZ_BYTES(query_hca_cap_out);
-	u32 in[MLX5_ST_SZ_DW(query_hca_cap_in)] = {};
-	void *out;
-	int err;
-
-	out = kzalloc(out_sz, GFP_KERNEL);
-	if (!out)
-		return -ENOMEM;
+	int err = mlx5_vport_get_vhca_id(mdev, vport, &info->vport_vhca_id);
 
-	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
-	MLX5_SET(query_hca_cap_in, in, other_function, true);
-	MLX5_SET(query_hca_cap_in, in, function_id, vport);
-	MLX5_SET(query_hca_cap_in, in, op_mod,
-		 MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE |
-		 HCA_CAP_OPMOD_GET_CUR);
-
-	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, out_sz);
 	if (err)
-		goto out;
-
-	info->vport_vhca_id = MLX5_GET(query_hca_cap_out, out,
-				       capability.cmd_hca_cap.vhca_id);
+		return err;
 
 	info->flags |= MLX5_IB_UAPI_QUERY_PORT_VPORT_VHCA_ID;
-out:
-	kfree(out);
-	return err;
+
+	return 0;
 }
 
 static int fill_multiport_info(struct mlx5_ib_dev *dev, u32 port_num,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/reporter_vnic.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/reporter_vnic.c
index 86253a89c24c..32bb769f1829 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/diag/reporter_vnic.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/reporter_vnic.c
@@ -1,6 +1,8 @@
 // SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
 /* Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. */
 
+#include <linux/mlx5/vport.h>
+
 #include "reporter_vnic.h"
 #include "en_stats.h"
 #include "devlink.h"
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
index b6d53db27cd5..81857c6f6bf7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
@@ -447,8 +447,6 @@ int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap
 #define mlx5_vport_get_other_func_general_cap(dev, vport, out)		\
 	mlx5_vport_get_other_func_cap(dev, vport, out, MLX5_CAP_GENERAL)
 
-int mlx5_vport_get_vhca_id(struct mlx5_core_dev *dev, u16 vport, u16 *vhca_id);
-
 static inline u32 mlx5_sriov_get_vf_total_msix(struct pci_dev *pdev)
 {
 	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/cmd.c
index 9c83753e4592..d447574d86fe 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/cmd.c
@@ -1199,22 +1199,28 @@ int mlx5hws_cmd_query_caps(struct mlx5_core_dev *mdev,
 int mlx5hws_cmd_query_gvmi(struct mlx5_core_dev *mdev, bool other_function,
 			   u16 vport_number, u16 *gvmi)
 {
-	bool ec_vf_func = other_function ? mlx5_core_is_ec_vf_vport(mdev, vport_number) : false;
 	u32 in[MLX5_ST_SZ_DW(query_hca_cap_in)] = {};
 	int out_size;
 	void *out;
 	int err;
 
+	if (other_function) {
+		err = mlx5_vport_get_vhca_id(mdev, vport_number, gvmi);
+		if  (!err)
+			return 0;
+
+		mlx5_core_err(mdev, "Failed to get vport vhca id for vport %d\n",
+			      vport_number);
+		return err;
+	}
+
+	/* get vhca_id for `this` function */
 	out_size = MLX5_ST_SZ_BYTES(query_hca_cap_out);
 	out = kzalloc(out_size, GFP_KERNEL);
 	if (!out)
 		return -ENOMEM;
 
 	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
-	MLX5_SET(query_hca_cap_in, in, other_function, other_function);
-	MLX5_SET(query_hca_cap_in, in, function_id,
-		 mlx5_vport_to_func_id(mdev, vport_number, ec_vf_func));
-	MLX5_SET(query_hca_cap_in, in, ec_vf_function, ec_vf_func);
 	MLX5_SET(query_hca_cap_in, in, op_mod,
 		 MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE << 1 | HCA_CAP_OPMOD_GET_CUR);
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_cmd.c
index baefb9a3fa05..bf99b933fd14 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_cmd.c
@@ -2,6 +2,7 @@
 /* Copyright (c) 2019 Mellanox Technologies. */
 
 #include "dr_types.h"
+#include "eswitch.h"
 
 int mlx5dr_cmd_query_esw_vport_context(struct mlx5_core_dev *mdev,
 				       bool other_vport,
@@ -34,21 +35,28 @@ int mlx5dr_cmd_query_esw_vport_context(struct mlx5_core_dev *mdev,
 int mlx5dr_cmd_query_gvmi(struct mlx5_core_dev *mdev, bool other_vport,
 			  u16 vport_number, u16 *gvmi)
 {
-	bool ec_vf_func = other_vport ? mlx5_core_is_ec_vf_vport(mdev, vport_number) : false;
 	u32 in[MLX5_ST_SZ_DW(query_hca_cap_in)] = {};
 	int out_size;
 	void *out;
 	int err;
 
+	if (other_vport) {
+		err = mlx5_vport_get_vhca_id(mdev, vport_number, gvmi);
+		if  (!err)
+			return 0;
+
+		mlx5_core_err(mdev, "Failed to get vport vhca id for vport %d\n",
+			      vport_number);
+		return err;
+	}
+
+	/* get vhca_id for `this` function */
 	out_size = MLX5_ST_SZ_BYTES(query_hca_cap_out);
 	out = kzalloc(out_size, GFP_KERNEL);
 	if (!out)
 		return -ENOMEM;
 
 	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
-	MLX5_SET(query_hca_cap_in, in, other_function, other_vport);
-	MLX5_SET(query_hca_cap_in, in, function_id, mlx5_vport_to_func_id(mdev, vport_number, ec_vf_func));
-	MLX5_SET(query_hca_cap_in, in, ec_vf_function, ec_vf_func);
 	MLX5_SET(query_hca_cap_in, in, op_mod,
 		 MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE << 1 |
 		 HCA_CAP_OPMOD_GET_CUR);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/vport.c b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
index 231bedc6a252..2ed2e530b07d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/vport.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
@@ -1239,7 +1239,9 @@ int mlx5_vport_get_vhca_id(struct mlx5_core_dev *dev, u16 vport, u16 *vhca_id)
 	void *hca_caps;
 	int err;
 
-	*vhca_id = 0;
+	/* try get vhca_id via eswitch */
+	if (mlx5_esw_vport_vhca_id(dev->priv.eswitch, vport, vhca_id))
+		return 0;
 
 	query_ctx = kzalloc(query_out_sz, GFP_KERNEL);
 	if (!query_ctx)
@@ -1256,6 +1258,7 @@ int mlx5_vport_get_vhca_id(struct mlx5_core_dev *dev, u16 vport, u16 *vhca_id)
 	kfree(query_ctx);
 	return err;
 }
+EXPORT_SYMBOL_GPL(mlx5_vport_get_vhca_id);
 
 int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap,
 				  u16 vport, u16 opmod)
diff --git a/include/linux/mlx5/vport.h b/include/linux/mlx5/vport.h
index c36cc6d82926..c87b9507cfa1 100644
--- a/include/linux/mlx5/vport.h
+++ b/include/linux/mlx5/vport.h
@@ -135,4 +135,6 @@ 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 vport, void *out,
 				  u16 opmod);
+int mlx5_vport_get_vhca_id(struct mlx5_core_dev *dev, u16 vport, u16 *vhca_id);
+
 #endif /* __MLX5_VPORT_H__ */
-- 
2.50.1

Re: [GIT PULL][mlx5-next 0/4] Cached vhca id and adjacent function vports

[Paolo Abeni]

On 8/15/25 9:48 PM, Saeed Mahameed wrote:
> From: Saeed Mahameed <saeedm@nvidia.com>
> 
> Hi Jakub, Jason,
> 
> This pull request introduces a preparation patchset for caching vhca_id 
> and needed HW bits for the upcoming netdev/eswitch series to support
> adjacent function vports.
> 
> For more information please see tag log below.

This has a conflict vs net-next, could you please double check the
conflict resolution?

Thanks,

Paolo

Re: [GIT PULL][mlx5-next 0/4] Cached vhca id and adjacent function vports

[Saeed Mahameed]

On 21 Aug 10:37, Paolo Abeni wrote:
>On 8/15/25 9:48 PM, Saeed Mahameed wrote:
>> From: Saeed Mahameed <saeedm@nvidia.com>
>>
>> Hi Jakub, Jason,
>>
>> This pull request introduces a preparation patchset for caching vhca_id
>> and needed HW bits for the upcoming netdev/eswitch series to support
>> adjacent function vports.
>>
>> For more information please see tag log below.
>
>This has a conflict vs net-next, could you please double check the
>conflict resolution?

Looks good, Thank you Paolo.
>
>Thanks,
>
>Paolo
>

Re: [GIT PULL][mlx5-next 0/4] Cached vhca id and adjacent function vports

[Jakub Kicinski]

On Fri, 15 Aug 2025 12:48:57 -0700 Saeed Mahameed wrote:
> To: Saeed Mahameed <saeedm@nvidia.com>, Leon Romanovsky <leonro@nvidia.com>

> Subject: [GIT PULL][mlx5-next 0/4] Cached vhca id and adjacent function vports

> Hi Jakub, Jason,

Sorry for not pulling this easier. FWIW, I completely missed your
intent, IDK if Jason read your mind right. The series is tagged *for*
mlx5-next, and To: yourself and Leon. It totally looked like you're
planning to apply this to your mlx5-next tree and send it out as a PR
later. I missed the "Hi Jakub, Jason" :$

If in doubt please don't put any tree designation in the subject
prefix.