net/mlx5e: Add devlink hairpin queues parameters

We refer to a TC NIC rule that involves forwarding as "hairpin".
Hairpin queues are mlx5 hardware specific implementation for hardware
forwarding of such packets.

Per the discussion in [1], move the hairpin queues control (number and
size) from debugfs to devlink.

Expose two devlink params:
- hairpin_num_queues: control the number of hairpin queues
- hairpin_queue_size: control the size (in packets) of the hairpin queues

[1] https://lore.kernel.org/all/[e-mail]/

Signed-off-by: Gal Pressman <[e-mail]>
Reviewed-by: Tariq Toukan <[e-mail]>
Signed-off-by: Saeed Mahameed <[e-mail]>
Link: https://lore.kernel.org/r/[e-mail]
Signed-off-by: Jakub Kicinski <[e-mail]>

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

···
 
     $ devlink dev eswitch set pci/0000:06:00.0 mode switchdev
 
+hairpin_num_queues: Number of hairpin queues
+--------------------------------------------
+We refer to a TC NIC rule that involves forwarding as "hairpin".
+
+Hairpin queues are mlx5 hardware specific implementation for hardware
+forwarding of such packets.
+
+- Show the number of hairpin queues::
+
+    $ devlink dev param show pci/0000:06:00.0 name hairpin_num_queues
+      pci/0000:06:00.0:
+        name hairpin_num_queues type driver-specific
+          values:
+            cmode driverinit value 2
+
+- Change the number of hairpin queues::
+
+    $ devlink dev param set pci/0000:06:00.0 name hairpin_num_queues value 4 cmode driverinit
+
+hairpin_queue_size: Size of the hairpin queues
+----------------------------------------------
+Control the size of the hairpin queues.
+
+- Show the size of the hairpin queues::
+
+    $ devlink dev param show pci/0000:06:00.0 name hairpin_queue_size
+      pci/0000:06:00.0:
+        name hairpin_queue_size type driver-specific
+          values:
+            cmode driverinit value 1024
+
+- Change the size (in packets) of the hairpin queues::
+
+    $ devlink dev param set pci/0000:06:00.0 name hairpin_queue_size value 512 cmode driverinit
+
 Health reporters
 ================
 

Documentation/networking/devlink/mlx5.rst

···
 
        Default: disabled
 
+   * - ``hairpin_num_queues``
+     - u32
+     - driverinit
+     - We refer to a TC NIC rule that involves forwarding as "hairpin".
+       Hairpin queues are mlx5 hardware specific implementation for hardware
+       forwarding of such packets.
+
+       Control the number of hairpin queues.
+   * - ``hairpin_queue_size``
+     - u32
+     - driverinit
+     - Control the size (in packets) of the hairpin queues.
 
 The ``mlx5`` driver supports reloading via ``DEVLINK_CMD_RELOAD``
 

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

···
 	return (val.vu32 >= 64 && val.vu32 <= 4096) ? 0 : -EINVAL;
 }
 
+static int
+mlx5_devlink_hairpin_num_queues_validate(struct devlink *devlink, u32 id,
+					 union devlink_param_value val,
+					 struct netlink_ext_ack *extack)
+{
+	return val.vu32 ? 0 : -EINVAL;
+}
+
+static int
+mlx5_devlink_hairpin_queue_size_validate(struct devlink *devlink, u32 id,
+					 union devlink_param_value val,
+					 struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	u32 val32 = val.vu32;
+
+	if (!is_power_of_2(val32)) {
+		NL_SET_ERR_MSG_MOD(extack, "Value is not power of two");
+		return -EINVAL;
+	}
+
+	if (val32 > BIT(MLX5_CAP_GEN(dev, log_max_hairpin_num_packets))) {
+		NL_SET_ERR_MSG_FMT_MOD(
+			extack, "Maximum hairpin queue size is %lu",
+			BIT(MLX5_CAP_GEN(dev, log_max_hairpin_num_packets)));
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void mlx5_devlink_hairpin_params_init_values(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	union devlink_param_value value;
+	u64 link_speed64;
+	u32 link_speed;
+
+	/* set hairpin pair per each 50Gbs share of the link */
+	mlx5_port_max_linkspeed(dev, &link_speed);
+	link_speed = max_t(u32, link_speed, 50000);
+	link_speed64 = link_speed;
+	do_div(link_speed64, 50000);
+
+	value.vu32 = link_speed64;
+	devl_param_driverinit_value_set(
+		devlink, MLX5_DEVLINK_PARAM_ID_HAIRPIN_NUM_QUEUES, value);
+
+	value.vu32 =
+		BIT(min_t(u32, 16 - MLX5_MPWRQ_MIN_LOG_STRIDE_SZ(dev),
+			  MLX5_CAP_GEN(dev, log_max_hairpin_num_packets)));
+	devl_param_driverinit_value_set(
+		devlink, MLX5_DEVLINK_PARAM_ID_HAIRPIN_QUEUE_SIZE, value);
+}
+
 static const struct devlink_param mlx5_devlink_params[] = {
 	DEVLINK_PARAM_GENERIC(ENABLE_ROCE, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
 			      NULL, NULL, mlx5_devlink_enable_roce_validate),
···
 static const struct devlink_param mlx5_devlink_eth_params[] = {
 	DEVLINK_PARAM_GENERIC(ENABLE_ETH, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
 			      NULL, NULL, NULL),
+	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_ID_HAIRPIN_NUM_QUEUES,
+			     "hairpin_num_queues", DEVLINK_PARAM_TYPE_U32,
+			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT), NULL, NULL,
+			     mlx5_devlink_hairpin_num_queues_validate),
+	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_ID_HAIRPIN_QUEUE_SIZE,
+			     "hairpin_queue_size", DEVLINK_PARAM_TYPE_U32,
+			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT), NULL, NULL,
+			     mlx5_devlink_hairpin_queue_size_validate),
 };
 
 static int mlx5_devlink_eth_params_register(struct devlink *devlink)
···
 	devl_param_driverinit_value_set(devlink,
 					DEVLINK_PARAM_GENERIC_ID_ENABLE_ETH,
 					value);
+
+	mlx5_devlink_hairpin_params_init_values(devlink);
+
 	return 0;
 }
 

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

···
 	MLX5_DEVLINK_PARAM_ID_ESW_LARGE_GROUP_NUM,
 	MLX5_DEVLINK_PARAM_ID_ESW_PORT_METADATA,
 	MLX5_DEVLINK_PARAM_ID_ESW_MULTIPORT,
+	MLX5_DEVLINK_PARAM_ID_HAIRPIN_NUM_QUEUES,
+	MLX5_DEVLINK_PARAM_ID_HAIRPIN_QUEUE_SIZE,
 };
 
 struct mlx5_trap_ctx {

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

···
 #include <net/bareudp.h>
 #include <net/bonding.h>
 #include <net/dst_metadata.h>
+#include "devlink.h"
 #include "en.h"
 #include "en/tc/post_act.h"
 #include "en/tc/act_stats.h"
···
 #define MLX5E_TC_TABLE_NUM_GROUPS 4
 #define MLX5E_TC_TABLE_MAX_GROUP_SIZE BIT(18)
 
-struct mlx5e_hairpin_params {
-	struct mlx5_core_dev *mdev;
-	u32 num_queues;
-	u32 queue_size;
-};
-
 struct mlx5e_tc_table {
 	/* Protects the dynamic assignment of the t parameter
 	 * which is the nic tc root table.
···
 
 	struct mlx5_tc_ct_priv         *ct;
 	struct mapping_ctx             *mapping;
-	struct mlx5e_hairpin_params    hairpin_params;
 	struct dentry                  *dfs_root;
 
 	/* tc action stats */
···
 			    &debugfs_hairpin_table_dump_fops);
 }
 
-static void
-mlx5e_hairpin_params_init(struct mlx5e_hairpin_params *hairpin_params,
-			  struct mlx5_core_dev *mdev)
-{
-	u64 link_speed64;
-	u32 link_speed;
-
-	hairpin_params->mdev = mdev;
-	/* set hairpin pair per each 50Gbs share of the link */
-	mlx5_port_max_linkspeed(mdev, &link_speed);
-	link_speed = max_t(u32, link_speed, 50000);
-	link_speed64 = link_speed;
-	do_div(link_speed64, 50000);
-	hairpin_params->num_queues = link_speed64;
-
-	hairpin_params->queue_size =
-		BIT(min_t(u32, 16 - MLX5_MPWRQ_MIN_LOG_STRIDE_SZ(mdev),
-			  MLX5_CAP_GEN(mdev, log_max_hairpin_num_packets)));
-}
-
 static int mlx5e_hairpin_flow_add(struct mlx5e_priv *priv,
 				  struct mlx5e_tc_flow *flow,
 				  struct mlx5e_tc_flow_parse_attr *parse_attr,
 				  struct netlink_ext_ack *extack)
 {
 	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct devlink *devlink = priv_to_devlink(priv->mdev);
 	int peer_ifindex = parse_attr->mirred_ifindex[0];
+	union devlink_param_value val = {};
 	struct mlx5_hairpin_params params;
 	struct mlx5_core_dev *peer_mdev;
 	struct mlx5e_hairpin_entry *hpe;
···
 		 hash_hairpin_info(peer_id, match_prio));
 	mutex_unlock(&tc->hairpin_tbl_lock);
 
-	params.log_num_packets = ilog2(tc->hairpin_params.queue_size);
+	err = devl_param_driverinit_value_get(
+		devlink, MLX5_DEVLINK_PARAM_ID_HAIRPIN_QUEUE_SIZE, &val);
+	if (err) {
+		err = -ENOMEM;
+		goto out_err;
+	}
+
+	params.log_num_packets = ilog2(val.vu32);
 	params.log_data_size =
 		clamp_t(u32,
 			params.log_num_packets +
···
 			MLX5_CAP_GEN(priv->mdev, log_max_hairpin_wq_data_sz));
 
 	params.q_counter = priv->q_counter;
-	params.num_channels = tc->hairpin_params.num_queues;
+	err = devl_param_driverinit_value_get(
+		devlink, MLX5_DEVLINK_PARAM_ID_HAIRPIN_NUM_QUEUES, &val);
+	if (err) {
+		err = -ENOMEM;
+		goto out_err;
+	}
+
+	params.num_channels = val.vu32;
 
 	hp = mlx5e_hairpin_create(priv, &params, peer_ifindex);
 	hpe->hp = hp;
···
 	tc->post_act = mlx5e_tc_post_act_init(priv, tc->chains, MLX5_FLOW_NAMESPACE_KERNEL);
 	tc->ct = mlx5_tc_ct_init(priv, tc->chains, &tc->mod_hdr,
 				 MLX5_FLOW_NAMESPACE_KERNEL, tc->post_act);
-
-	mlx5e_hairpin_params_init(&tc->hairpin_params, dev);
 
 	tc->netdevice_nb.notifier_call = mlx5e_tc_netdev_event;
 	err = register_netdevice_notifier_dev_net(priv->netdev,