[pull request][net-next 00/15] mlx5 updates 2023-02-10

[pull request][net-next 00/15] mlx5 updates 2023-02-10

[Saeed Mahameed]

From: Saeed Mahameed <saeedm@nvidia.com>

This series adds multip port eswitch support and some devlink and
auxiliary device management improvements in mlx5.

For more information please see tag log below.

Please pull and let me know if there is any problem.

Thanks,
Saeed.


The following changes since commit 025a785ff083729819dc82ac81baf190cb4aee5c:

  net: skbuff: drop the word head from skb cache (2023-02-10 09:10:28 +0000)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-updates-2023-02-10

for you to fetch changes up to c1328578042eaa5d3f279ed636160dbac660daaa:

  net/mlx5: Suspend auxiliary devices only in case of PCI device suspend (2023-02-10 14:16:29 -0800)

----------------------------------------------------------------
mlx5-updates-2023-02-10

1) From Roi and Mark: MultiPort eswitch support

MultiPort E-Switch builds on newer hardware's capabilities and introduces
a mode where a single E-Switch is used and all the vports and physical
ports on the NIC are connected to it.

The new mode will allow in the future a decrease in the memory used by the
driver and advanced features that aren't possible today.

This represents a big change in the current E-Switch implantation in mlx5.
Currently, by default, each E-Switch manager manages its E-Switch.
Steering rules in each E-Switch can only forward traffic to the native
physical port associated with that E-Switch. While there are ways to target
non-native physical ports, for example using a bond or via special TC
rules. None of the ways allows a user to configure the driver
to operate by default in such a mode nor can the driver decide
to move to this mode by default as it's user configuration-driven right now.

While MultiPort E-Switch single FDB mode is the preferred mode, older
generations of ConnectX hardware couldn't support this mode so it was never
implemented. Now that there is capable hardware present, start the
transition to having this mode by default.

Introduce a devlink parameter to control MultiPort Eswitch single FDB mode.
This will allow users to select this mode on their system right now
and in the future will allow the driver to move to this mode by default.

2) From Jiri: Improvements and fixes for mlx5 netdev's devlink logic
 2.1) Cleanups related to mlx5's devlink port logic
 2.2) Move devlink port registration to be done before netdev alloc
 2.3) Create auxdev devlink instance in the same ns as parent devlink
 2.4) Suspend auxiliary devices only in case of PCI device suspend

----------------------------------------------------------------
Jiri Pirko (8):
      net/mlx5: Remove outdated comment
      net/mlx5e: Pass mdev to mlx5e_devlink_port_register()
      net/mlx5e: Replace usage of mlx5e_devlink_get_dl_port() by netdev->devlink_port
      net/mlx5e: Move dl_port to struct mlx5e_dev
      net/mlx5e: Move devlink port registration to be done before netdev alloc
      net/mlx5e: Create auxdev devlink instance in the same ns as parent devlink
      net/mlx5: Remove "recovery" arg from mlx5_load_one() function
      net/mlx5: Suspend auxiliary devices only in case of PCI device suspend

Mark Bloch (1):
      net/mlx5: Lag, Add single RDMA device in multiport mode

Roi Dayan (6):
      net/mlx5: Lag, Let user configure multiport eswitch
      net/mlx5e: TC, Add peer flow in mpesw mode
      net/mlx5: E-Switch, rename bond update function to be reused
      net/mlx5: Lag, set different uplink vport metadata in multiport eswitch mode
      net/mlx5e: Use a simpler comparison for uplink rep
      net/mlx5e: TC, Remove redundant parse_attr argument

 Documentation/networking/devlink/mlx5.rst          |   4 +
 drivers/infiniband/hw/mlx5/ib_rep.c                |  18 ++-
 drivers/net/ethernet/mellanox/mlx5/core/dev.c      |   8 +-
 drivers/net/ethernet/mellanox/mlx5/core/devlink.c  |  60 ++++++++-
 drivers/net/ethernet/mellanox/mlx5/core/devlink.h  |   1 +
 drivers/net/ethernet/mellanox/mlx5/core/en.h       |   1 +
 .../net/ethernet/mellanox/mlx5/core/en/devlink.c   |  34 +++--
 .../net/ethernet/mellanox/mlx5/core/en/devlink.h   |  13 +-
 .../net/ethernet/mellanox/mlx5/core/en/rep/bond.c  |   6 +-
 .../ethernet/mellanox/mlx5/core/en/reporter_rx.c   |   4 +-
 .../ethernet/mellanox/mlx5/core/en/reporter_tx.c   |   4 +-
 .../ethernet/mellanox/mlx5/core/en/tc/act/mirred.c |   9 --
 .../net/ethernet/mellanox/mlx5/core/en/tc_tun.c    |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c  |  26 ++--
 drivers/net/ethernet/mellanox/mlx5/core/en_rep.c   |  35 +++++
 drivers/net/ethernet/mellanox/mlx5/core/en_rep.h   |   1 +
 drivers/net/ethernet/mellanox/mlx5/core/en_rx.c    |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c    |  32 ++---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.h    |   6 -
 .../mellanox/mlx5/core/esw/acl/ingress_ofld.c      |   4 +-
 .../net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h |   4 +-
 .../ethernet/mellanox/mlx5/core/eswitch_offloads.c |   2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/health.c   |   2 +-
 drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c  |  10 +-
 drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h  |   4 +
 .../net/ethernet/mellanox/mlx5/core/lag/mpesw.c    | 144 ++++++++++++++++-----
 .../net/ethernet/mellanox/mlx5/core/lag/mpesw.h    |  15 +--
 drivers/net/ethernet/mellanox/mlx5/core/main.c     |  25 ++--
 .../net/ethernet/mellanox/mlx5/core/mlx5_core.h    |   8 +-
 .../ethernet/mellanox/mlx5/core/sf/dev/driver.c    |   2 +-
 include/linux/mlx5/driver.h                        |   3 +-
 32 files changed, 321 insertions(+), 182 deletions(-)

[net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Saeed Mahameed]

From: Roi Dayan <roid@nvidia.com>

Instead of activating multiport eswitch dynamically through
adding a TC rule and meeting certain conditions, allow the user
to activate it through devlink.
This will remove the forced requirement of using TC.
e.g. Bridge offload.

Example:
    $ devlink dev param set pci/0000:00:0b.0 name esw_multiport value 1 \
                  cmode runtime

Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 Documentation/networking/devlink/mlx5.rst     |  4 ++
 .../net/ethernet/mellanox/mlx5/core/devlink.c | 56 +++++++++++++++++++
 .../net/ethernet/mellanox/mlx5/core/devlink.h |  1 +
 .../mellanox/mlx5/core/en/tc/act/mirred.c     |  9 ---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 22 +-------
 .../net/ethernet/mellanox/mlx5/core/en_tc.h   |  6 --
 .../net/ethernet/mellanox/mlx5/core/lag/lag.c |  4 +-
 .../net/ethernet/mellanox/mlx5/core/lag/lag.h |  1 +
 .../ethernet/mellanox/mlx5/core/lag/mpesw.c   | 46 +++++++--------
 .../ethernet/mellanox/mlx5/core/lag/mpesw.h   | 12 +---
 10 files changed, 87 insertions(+), 74 deletions(-)

diff --git a/Documentation/networking/devlink/mlx5.rst b/Documentation/networking/devlink/mlx5.rst
index 29ad304e6fba..1d2ad2727da1 100644
--- a/Documentation/networking/devlink/mlx5.rst
+++ b/Documentation/networking/devlink/mlx5.rst
@@ -54,6 +54,10 @@ parameters.
      - Control the number of large groups (size > 1) in the FDB table.
 
        * The default value is 15, and the range is between 1 and 1024.
+   * - ``esw_multiport``
+     - Boolean
+     - runtime
+     - Set the E-Switch lag mode to multiport.
 
 The ``mlx5`` driver supports reloading via ``DEVLINK_CMD_RELOAD``
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
index b742e04deec1..49392870f695 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
@@ -7,6 +7,7 @@
 #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"
@@ -437,6 +438,55 @@ static int mlx5_devlink_large_group_num_validate(struct devlink *devlink, u32 id
 	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);
+	int err = 0;
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return -EOPNOTSUPP;
+
+	if (ctx->val.vbool)
+		err = mlx5_lag_mpesw_enable(dev);
+	else
+		mlx5_lag_mpesw_disable(dev);
+
+	return err;
+}
+
+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_mpesw_is_activated(dev);
+	return 0;
+}
+
+static int mlx5_devlink_esw_multiport_validate(struct devlink *devlink, u32 id,
+					       union devlink_param_value val,
+					       struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (!MLX5_ESWITCH_MANAGER(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch is unsupported");
+		return -EOPNOTSUPP;
+	}
+
+	if (mlx5_eswitch_mode(dev) != MLX5_ESWITCH_OFFLOADS) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "E-Switch must be in switchdev mode");
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
 #endif
 
 static int mlx5_devlink_eq_depth_validate(struct devlink *devlink, u32 id,
@@ -455,6 +505,12 @@ static const struct devlink_param mlx5_devlink_params[] = {
 			     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,
+			     mlx5_devlink_esw_multiport_validate),
 #endif
 	DEVLINK_PARAM_GENERIC(IO_EQ_SIZE, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
 			      NULL, NULL, mlx5_devlink_eq_depth_validate),
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.h b/drivers/net/ethernet/mellanox/mlx5/core/devlink.h
index b561107e0df1..212b12424146 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/devlink.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.h
@@ -11,6 +11,7 @@ enum mlx5_devlink_param_id {
 	MLX5_DEVLINK_PARAM_ID_FLOW_STEERING_MODE,
 	MLX5_DEVLINK_PARAM_ID_ESW_LARGE_GROUP_NUM,
 	MLX5_DEVLINK_PARAM_ID_ESW_PORT_METADATA,
+	MLX5_DEVLINK_PARAM_ID_ESW_MULTIPORT,
 };
 
 struct mlx5_trap_ctx {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred.c
index c095a12346de..07cc65596f89 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred.c
@@ -216,7 +216,6 @@ parse_mirred(struct mlx5e_tc_act_parse_state *parse_state,
 	struct net_device *uplink_dev;
 	struct mlx5e_priv *out_priv;
 	struct mlx5_eswitch *esw;
-	bool is_uplink_rep;
 	int *ifindexes;
 	int if_count;
 	int err;
@@ -231,7 +230,6 @@ parse_mirred(struct mlx5e_tc_act_parse_state *parse_state,
 
 	parse_state->ifindexes[if_count] = out_dev->ifindex;
 	parse_state->if_count++;
-	is_uplink_rep = mlx5e_eswitch_uplink_rep(out_dev);
 
 	if (mlx5_lag_mpesw_do_mirred(priv->mdev, out_dev, extack))
 		return -EOPNOTSUPP;
@@ -275,13 +273,6 @@ parse_mirred(struct mlx5e_tc_act_parse_state *parse_state,
 	esw_attr->dests[esw_attr->out_count].rep = rpriv->rep;
 	esw_attr->dests[esw_attr->out_count].mdev = out_priv->mdev;
 
-	/* If output device is bond master then rules are not explicit
-	 * so we don't attempt to count them.
-	 */
-	if (is_uplink_rep && MLX5_CAP_PORT_SELECTION(priv->mdev, port_select_flow_table) &&
-	    MLX5_CAP_GEN(priv->mdev, create_lag_when_not_master_up))
-		attr->lag.count = true;
-
 	esw_attr->out_count++;
 
 	return 0;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index e2ec80ebde58..00d5b0aa295b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -2124,9 +2124,6 @@ static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
 	free_branch_attr(flow, attr->branch_true);
 	free_branch_attr(flow, attr->branch_false);
 
-	if (flow->attr->lag.count)
-		mlx5_lag_del_mpesw_rule(esw->dev);
-
 	kvfree(attr->esw_attr->rx_tun_attr);
 	kvfree(attr->parse_attr);
 	kfree(flow->attr);
@@ -4277,12 +4274,7 @@ static bool is_lag_dev(struct mlx5e_priv *priv,
 
 static bool is_multiport_eligible(struct mlx5e_priv *priv, struct net_device *out_dev)
 {
-	if (same_hw_reps(priv, out_dev) &&
-	    MLX5_CAP_PORT_SELECTION(priv->mdev, port_select_flow_table) &&
-	    MLX5_CAP_GEN(priv->mdev, create_lag_when_not_master_up))
-		return true;
-
-	return false;
+	return same_hw_reps(priv, out_dev) && mlx5_lag_mpesw_is_activated(priv->mdev);
 }
 
 bool mlx5e_is_valid_eswitch_fwd_dev(struct mlx5e_priv *priv,
@@ -4584,7 +4576,6 @@ __mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
 		     struct mlx5_core_dev *in_mdev)
 {
 	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
-	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct netlink_ext_ack *extack = f->common.extack;
 	struct mlx5e_tc_flow_parse_attr *parse_attr;
 	struct mlx5e_tc_flow *flow;
@@ -4617,26 +4608,17 @@ __mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
 	if (err)
 		goto err_free;
 
-	if (flow->attr->lag.count) {
-		err = mlx5_lag_add_mpesw_rule(esw->dev);
-		if (err)
-			goto err_free;
-	}
-
 	err = mlx5e_tc_add_fdb_flow(priv, flow, extack);
 	complete_all(&flow->init_done);
 	if (err) {
 		if (!(err == -ENETUNREACH && mlx5_lag_is_multipath(in_mdev)))
-			goto err_lag;
+			goto err_free;
 
 		add_unready_flow(flow);
 	}
 
 	return flow;
 
-err_lag:
-	if (flow->attr->lag.count)
-		mlx5_lag_del_mpesw_rule(esw->dev);
 err_free:
 	mlx5e_flow_put(priv, flow);
 out:
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.h b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.h
index ce516dc7f3fd..a0ed9d95a634 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.h
@@ -90,12 +90,6 @@ struct mlx5_flow_attr {
 	u32 exe_aso_type;
 	struct list_head list;
 	struct mlx5e_post_act_handle *post_act_handle;
-	struct {
-		/* Indicate whether the parsed flow should be counted for lag mode decision
-		 * making
-		 */
-		bool count;
-	} lag;
 	struct mlx5_flow_attr *branch_true;
 	struct mlx5_flow_attr *branch_false;
 	struct mlx5_flow_attr *jumping_attr;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
index dbf218cac535..301994741b08 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
@@ -230,7 +230,6 @@ static void mlx5_ldev_free(struct kref *ref)
 	mlx5_lag_mp_cleanup(ldev);
 	cancel_delayed_work_sync(&ldev->bond_work);
 	destroy_workqueue(ldev->wq);
-	mlx5_lag_mpesw_cleanup(ldev);
 	mutex_destroy(&ldev->lock);
 	kfree(ldev);
 }
@@ -276,7 +275,6 @@ static struct mlx5_lag *mlx5_lag_dev_alloc(struct mlx5_core_dev *dev)
 		mlx5_core_err(dev, "Failed to init multipath lag err=%d\n",
 			      err);
 
-	mlx5_lag_mpesw_init(ldev);
 	ldev->ports = MLX5_CAP_GEN(dev, num_lag_ports);
 	ldev->buckets = 1;
 
@@ -688,7 +686,7 @@ static int mlx5_deactivate_lag(struct mlx5_lag *ldev)
 }
 
 #define MLX5_LAG_OFFLOADS_SUPPORTED_PORTS 2
-static bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
+bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
 {
 #ifdef CONFIG_MLX5_ESWITCH
 	struct mlx5_core_dev *dev;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
index 66013bef9939..2dbd96a86ef8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
@@ -102,6 +102,7 @@ mlx5_lag_is_ready(struct mlx5_lag *ldev)
 	return test_bit(MLX5_LAG_FLAG_NDEVS_READY, &ldev->state_flags);
 }
 
+bool mlx5_lag_check_prereq(struct mlx5_lag *ldev);
 void mlx5_modify_lag(struct mlx5_lag *ldev,
 		     struct lag_tracker *tracker);
 int mlx5_activate_lag(struct mlx5_lag *ldev,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index 3799f89ed1a6..3f8fc965cec6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -7,18 +7,19 @@
 #include "eswitch.h"
 #include "lib/mlx5.h"
 
-static int add_mpesw_rule(struct mlx5_lag *ldev)
+static int enable_mpesw(struct mlx5_lag *ldev)
 {
 	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
 	int err;
 
-	if (atomic_add_return(1, &ldev->lag_mpesw.mpesw_rule_count) != 1)
-		return 0;
+	if (ldev->mode != MLX5_LAG_MODE_NONE)
+		return -EINVAL;
 
-	if (ldev->mode != MLX5_LAG_MODE_NONE) {
-		err = -EINVAL;
-		goto out_err;
-	}
+	if (mlx5_eswitch_mode(dev) != MLX5_ESWITCH_OFFLOADS ||
+	    !MLX5_CAP_PORT_SELECTION(dev, port_select_flow_table) ||
+	    !MLX5_CAP_GEN(dev, create_lag_when_not_master_up) ||
+	    !mlx5_lag_check_prereq(ldev))
+		return -EOPNOTSUPP;
 
 	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, false);
 	if (err) {
@@ -29,14 +30,12 @@ static int add_mpesw_rule(struct mlx5_lag *ldev)
 	return 0;
 
 out_err:
-	atomic_dec(&ldev->lag_mpesw.mpesw_rule_count);
 	return err;
 }
 
-static void del_mpesw_rule(struct mlx5_lag *ldev)
+static void disable_mpesw(struct mlx5_lag *ldev)
 {
-	if (!atomic_dec_return(&ldev->lag_mpesw.mpesw_rule_count) &&
-	    ldev->mode == MLX5_LAG_MODE_MPESW)
+	if (ldev->mode == MLX5_LAG_MODE_MPESW)
 		mlx5_disable_lag(ldev);
 }
 
@@ -46,12 +45,17 @@ static void mlx5_mpesw_work(struct work_struct *work)
 	struct mlx5_lag *ldev = mpesww->lag;
 
 	mutex_lock(&ldev->lock);
+	if (ldev->mode_changes_in_progress) {
+		mpesww->result = -EAGAIN;
+		goto unlock;
+	}
+
 	if (mpesww->op == MLX5_MPESW_OP_ENABLE)
-		mpesww->result = add_mpesw_rule(ldev);
+		mpesww->result = enable_mpesw(ldev);
 	else if (mpesww->op == MLX5_MPESW_OP_DISABLE)
-		del_mpesw_rule(ldev);
+		disable_mpesw(ldev);
+unlock:
 	mutex_unlock(&ldev->lock);
-
 	complete(&mpesww->comp);
 }
 
@@ -86,12 +90,12 @@ static int mlx5_lag_mpesw_queue_work(struct mlx5_core_dev *dev,
 	return err;
 }
 
-void mlx5_lag_del_mpesw_rule(struct mlx5_core_dev *dev)
+void mlx5_lag_mpesw_disable(struct mlx5_core_dev *dev)
 {
 	mlx5_lag_mpesw_queue_work(dev, MLX5_MPESW_OP_DISABLE);
 }
 
-int mlx5_lag_add_mpesw_rule(struct mlx5_core_dev *dev)
+int mlx5_lag_mpesw_enable(struct mlx5_core_dev *dev)
 {
 	return mlx5_lag_mpesw_queue_work(dev, MLX5_MPESW_OP_ENABLE);
 }
@@ -118,13 +122,3 @@ bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev)
 
 	return ldev && ldev->mode == MLX5_LAG_MODE_MPESW;
 }
-
-void mlx5_lag_mpesw_init(struct mlx5_lag *ldev)
-{
-	atomic_set(&ldev->lag_mpesw.mpesw_rule_count, 0);
-}
-
-void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev)
-{
-	WARN_ON(atomic_read(&ldev->lag_mpesw.mpesw_rule_count));
-}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
index 818f19b5a984..571e4acf262e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
@@ -9,7 +9,6 @@
 
 struct lag_mpesw {
 	struct work_struct mpesw_work;
-	atomic_t mpesw_rule_count;
 };
 
 enum mpesw_op {
@@ -29,14 +28,7 @@ int mlx5_lag_mpesw_do_mirred(struct mlx5_core_dev *mdev,
 			     struct net_device *out_dev,
 			     struct netlink_ext_ack *extack);
 bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev);
-void mlx5_lag_del_mpesw_rule(struct mlx5_core_dev *dev);
-int mlx5_lag_add_mpesw_rule(struct mlx5_core_dev *dev);
-#if IS_ENABLED(CONFIG_MLX5_ESWITCH)
-void mlx5_lag_mpesw_init(struct mlx5_lag *ldev);
-void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev);
-#else
-static inline void mlx5_lag_mpesw_init(struct mlx5_lag *ldev) {}
-static inline void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev) {}
-#endif
+void mlx5_lag_mpesw_disable(struct mlx5_core_dev *dev);
+int mlx5_lag_mpesw_enable(struct mlx5_core_dev *dev);
 
 #endif /* __MLX5_LAG_MPESW_H__ */
-- 
2.39.1

[net-next 02/15] net/mlx5e: TC, Add peer flow in mpesw mode

[Saeed Mahameed]

From: Roi Dayan <roid@nvidia.com>

While at it rename mlx5_lag_mpesw_is_activated() to mlx5_lag_is_mpesw() to
be consistent with checking if other lag modes are activated.

Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/devlink.c        | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c      | 3 ++-
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c          | 9 +++++++--
 .../net/ethernet/mellanox/mlx5/core/eswitch_offloads.c   | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c      | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h      | 2 +-
 6 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
index 49392870f695..1bccc6c31460 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
@@ -463,7 +463,7 @@ static int mlx5_devlink_esw_multiport_get(struct devlink *devlink, u32 id,
 	if (!MLX5_ESWITCH_MANAGER(dev))
 		return -EOPNOTSUPP;
 
-	ctx->val.vbool = mlx5_lag_mpesw_is_activated(dev);
+	ctx->val.vbool = mlx5_lag_is_mpesw(dev);
 	return 0;
 }
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
index 83bb0811e774..684c0293a4d6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
@@ -97,7 +97,8 @@ static int get_route_and_out_devs(struct mlx5e_priv *priv,
 	      mlx5e_is_uplink_rep(netdev_priv(*out_dev))))
 		return -EOPNOTSUPP;
 
-	if (mlx5e_eswitch_uplink_rep(priv->netdev) && *out_dev != priv->netdev)
+	if (mlx5e_eswitch_uplink_rep(priv->netdev) && *out_dev != priv->netdev &&
+	    !mlx5_lag_is_mpesw(priv->mdev))
 		return -EOPNOTSUPP;
 
 	return 0;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 00d5b0aa295b..780b2aa2ace1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -4274,7 +4274,7 @@ static bool is_lag_dev(struct mlx5e_priv *priv,
 
 static bool is_multiport_eligible(struct mlx5e_priv *priv, struct net_device *out_dev)
 {
-	return same_hw_reps(priv, out_dev) && mlx5_lag_mpesw_is_activated(priv->mdev);
+	return same_hw_reps(priv, out_dev) && mlx5_lag_is_mpesw(priv->mdev);
 }
 
 bool mlx5e_is_valid_eswitch_fwd_dev(struct mlx5e_priv *priv,
@@ -4445,6 +4445,9 @@ static bool is_peer_flow_needed(struct mlx5e_tc_flow *flow)
 	    (is_rep_ingress || act_is_encap))
 		return true;
 
+	if (mlx5_lag_is_mpesw(esw_attr->in_mdev))
+		return true;
+
 	return false;
 }
 
@@ -4650,8 +4653,10 @@ static int mlx5e_tc_add_fdb_peer_flow(struct flow_cls_offload *f,
 	 * So packets redirected to uplink use the same mdev of the
 	 * original flow and packets redirected from uplink use the
 	 * peer mdev.
+	 * In multiport eswitch it's a special case that we need to
+	 * keep the original mdev.
 	 */
-	if (attr->in_rep->vport == MLX5_VPORT_UPLINK)
+	if (attr->in_rep->vport == MLX5_VPORT_UPLINK && !mlx5_lag_is_mpesw(priv->mdev))
 		in_mdev = peer_priv->mdev;
 	else
 		in_mdev = priv->mdev;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 8fb09143e9e8..2a98375a0abf 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -443,7 +443,7 @@ esw_setup_vport_dest(struct mlx5_flow_destination *dest, struct mlx5_flow_act *f
 			MLX5_CAP_GEN(esw_attr->dests[attr_idx].mdev, vhca_id);
 		dest[dest_idx].vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
 		if (dest[dest_idx].vport.num == MLX5_VPORT_UPLINK &&
-		    mlx5_lag_mpesw_is_activated(esw->dev))
+		    mlx5_lag_is_mpesw(esw->dev))
 			dest[dest_idx].type = MLX5_FLOW_DESTINATION_TYPE_UPLINK;
 	}
 	if (esw_attr->dests[attr_idx].flags & MLX5_ESW_DEST_ENCAP_VALID) {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index 3f8fc965cec6..dd3cb9aa06fd 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -116,7 +116,7 @@ int mlx5_lag_mpesw_do_mirred(struct mlx5_core_dev *mdev,
 	return -EOPNOTSUPP;
 }
 
-bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev)
+bool mlx5_lag_is_mpesw(struct mlx5_core_dev *dev)
 {
 	struct mlx5_lag *ldev = mlx5_lag_dev(dev);
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
index 571e4acf262e..d857ea988bf2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
@@ -27,7 +27,7 @@ struct mlx5_mpesw_work_st {
 int mlx5_lag_mpesw_do_mirred(struct mlx5_core_dev *mdev,
 			     struct net_device *out_dev,
 			     struct netlink_ext_ack *extack);
-bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev);
+bool mlx5_lag_is_mpesw(struct mlx5_core_dev *dev);
 void mlx5_lag_mpesw_disable(struct mlx5_core_dev *dev);
 int mlx5_lag_mpesw_enable(struct mlx5_core_dev *dev);
 
-- 
2.39.1

[net-next 03/15] net/mlx5: E-Switch, rename bond update function to be reused

[Saeed Mahameed]

From: Roi Dayan <roid@nvidia.com>

The vport bond update function is really updating the vport metadata
and there is no direct relation to bond. Rename the function
to vport metadata update to be used a followup commit.
This commit doesn't change any functionality.

Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c       | 6 +++---
 .../net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c  | 4 ++--
 drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h      | 4 ++--
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c
index b6f5c1bcdbcd..016a61c52c45 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c
@@ -120,8 +120,8 @@ int mlx5e_rep_bond_enslave(struct mlx5_eswitch *esw, struct net_device *netdev,
 	priv = netdev_priv(netdev);
 	rpriv = priv->ppriv;
 
-	err = mlx5_esw_acl_ingress_vport_bond_update(esw, rpriv->rep->vport,
-						     mdata->metadata_reg_c_0);
+	err = mlx5_esw_acl_ingress_vport_metadata_update(esw, rpriv->rep->vport,
+							 mdata->metadata_reg_c_0);
 	if (err)
 		goto ingress_err;
 
@@ -167,7 +167,7 @@ void mlx5e_rep_bond_unslave(struct mlx5_eswitch *esw,
 	/* Reset bond_metadata to zero first then reset all ingress/egress
 	 * acls and rx rules of unslave representor's vport
 	 */
-	mlx5_esw_acl_ingress_vport_bond_update(esw, rpriv->rep->vport, 0);
+	mlx5_esw_acl_ingress_vport_metadata_update(esw, rpriv->rep->vport, 0);
 	mlx5_esw_acl_egress_vport_unbond(esw, rpriv->rep->vport);
 	mlx5e_rep_bond_update(priv, false);
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
index a994e71e05c1..d55775627a47 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
@@ -356,8 +356,8 @@ void esw_acl_ingress_ofld_cleanup(struct mlx5_eswitch *esw,
 }
 
 /* Caller must hold rtnl_lock */
-int mlx5_esw_acl_ingress_vport_bond_update(struct mlx5_eswitch *esw, u16 vport_num,
-					   u32 metadata)
+int mlx5_esw_acl_ingress_vport_metadata_update(struct mlx5_eswitch *esw, u16 vport_num,
+					       u32 metadata)
 {
 	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
 	int err;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h
index 11d3d3978848..c9f8469e9a47 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h
@@ -24,8 +24,8 @@ static inline bool mlx5_esw_acl_egress_fwd2vport_supported(struct mlx5_eswitch *
 /* Eswitch acl ingress external APIs */
 int esw_acl_ingress_ofld_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
 void esw_acl_ingress_ofld_cleanup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
-int mlx5_esw_acl_ingress_vport_bond_update(struct mlx5_eswitch *esw, u16 vport_num,
-					   u32 metadata);
+int mlx5_esw_acl_ingress_vport_metadata_update(struct mlx5_eswitch *esw, u16 vport_num,
+					       u32 metadata);
 void mlx5_esw_acl_ingress_vport_drop_rule_destroy(struct mlx5_eswitch *esw, u16 vport_num);
 int mlx5_esw_acl_ingress_vport_drop_rule_create(struct mlx5_eswitch *esw, u16 vport_num);
 
-- 
2.39.1

[net-next 04/15] net/mlx5: Lag, set different uplink vport metadata in multiport eswitch mode

[Saeed Mahameed]

From: Roi Dayan <roid@nvidia.com>

In a follow-up commit multiport eswitch mode will use a shared fdb.
In shared fdb there is a single eswitch fdb and traffic could come from any
port. to distinguish between the ports set a different metadata per uplink port.

Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_rep.c  | 35 ++++++++++
 .../net/ethernet/mellanox/mlx5/core/en_rep.h  |  1 +
 .../ethernet/mellanox/mlx5/core/lag/mpesw.c   | 67 ++++++++++++++++++-
 .../ethernet/mellanox/mlx5/core/lag/mpesw.h   |  1 +
 include/linux/mlx5/driver.h                   |  1 +
 5 files changed, 104 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
index 8d29310c7e48..9b9203443085 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
@@ -1007,8 +1007,23 @@ static void mlx5e_cleanup_rep_rx(struct mlx5e_priv *priv)
 	priv->rx_res = NULL;
 }
 
+static void mlx5e_rep_mpesw_work(struct work_struct *work)
+{
+	struct mlx5_rep_uplink_priv *uplink_priv =
+		container_of(work, struct mlx5_rep_uplink_priv,
+			     mpesw_work);
+	struct mlx5e_rep_priv *rpriv =
+		container_of(uplink_priv, struct mlx5e_rep_priv,
+			     uplink_priv);
+	struct mlx5e_priv *priv = netdev_priv(rpriv->netdev);
+
+	rep_vport_rx_rule_destroy(priv);
+	mlx5e_create_rep_vport_rx_rule(priv);
+}
+
 static int mlx5e_init_ul_rep_rx(struct mlx5e_priv *priv)
 {
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
 	int err;
 
 	mlx5e_create_q_counters(priv);
@@ -1018,12 +1033,17 @@ static int mlx5e_init_ul_rep_rx(struct mlx5e_priv *priv)
 
 	mlx5e_tc_int_port_init_rep_rx(priv);
 
+	INIT_WORK(&rpriv->uplink_priv.mpesw_work, mlx5e_rep_mpesw_work);
+
 out:
 	return err;
 }
 
 static void mlx5e_cleanup_ul_rep_rx(struct mlx5e_priv *priv)
 {
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+
+	cancel_work_sync(&rpriv->uplink_priv.mpesw_work);
 	mlx5e_tc_int_port_cleanup_rep_rx(priv);
 	mlx5e_cleanup_rep_rx(priv);
 	mlx5e_destroy_q_counters(priv);
@@ -1132,6 +1152,19 @@ static int mlx5e_update_rep_rx(struct mlx5e_priv *priv)
 	return 0;
 }
 
+static int mlx5e_rep_event_mpesw(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+
+	if (rep->vport != MLX5_VPORT_UPLINK)
+		return NOTIFY_DONE;
+
+	queue_work(priv->wq, &rpriv->uplink_priv.mpesw_work);
+
+	return NOTIFY_OK;
+}
+
 static int uplink_rep_async_event(struct notifier_block *nb, unsigned long event, void *data)
 {
 	struct mlx5e_priv *priv = container_of(nb, struct mlx5e_priv, events_nb);
@@ -1153,6 +1186,8 @@ static int uplink_rep_async_event(struct notifier_block *nb, unsigned long event
 
 	if (event == MLX5_DEV_EVENT_PORT_AFFINITY)
 		return mlx5e_rep_tc_event_port_affinity(priv);
+	else if (event == MLX5_DEV_EVENT_MULTIPORT_ESW)
+		return mlx5e_rep_event_mpesw(priv);
 
 	return NOTIFY_DONE;
 }
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
index b4e691760da9..a0891caf464e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
@@ -100,6 +100,7 @@ struct mlx5_rep_uplink_priv {
 	struct mlx5e_tc_int_port_priv *int_port_priv;
 
 	struct mlx5e_flow_meters *flow_meters;
+	struct work_struct mpesw_work;
 };
 
 struct mlx5e_rep_priv {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index dd3cb9aa06fd..2f7f2af312d7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -5,8 +5,66 @@
 #include <net/nexthop.h>
 #include "lag/lag.h"
 #include "eswitch.h"
+#include "esw/acl/ofld.h"
 #include "lib/mlx5.h"
 
+static void mlx5_mpesw_metadata_cleanup(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev;
+	struct mlx5_eswitch *esw;
+	u32 pf_metadata;
+	int i;
+
+	for (i = 0; i < ldev->ports; i++) {
+		dev = ldev->pf[i].dev;
+		esw = dev->priv.eswitch;
+		pf_metadata = ldev->lag_mpesw.pf_metadata[i];
+		if (!pf_metadata)
+			continue;
+		mlx5_esw_acl_ingress_vport_metadata_update(esw, MLX5_VPORT_UPLINK, 0);
+		mlx5_notifier_call_chain(dev->priv.events, MLX5_DEV_EVENT_MULTIPORT_ESW,
+					 (void *)0);
+		mlx5_esw_match_metadata_free(esw, pf_metadata);
+		ldev->lag_mpesw.pf_metadata[i] = 0;
+	}
+}
+
+static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev;
+	struct mlx5_eswitch *esw;
+	u32 pf_metadata;
+	int i, err;
+
+	for (i = 0; i < ldev->ports; i++) {
+		dev = ldev->pf[i].dev;
+		esw = dev->priv.eswitch;
+		pf_metadata = mlx5_esw_match_metadata_alloc(esw);
+		if (!pf_metadata) {
+			err = -ENOSPC;
+			goto err_metadata;
+		}
+
+		ldev->lag_mpesw.pf_metadata[i] = pf_metadata;
+		err = mlx5_esw_acl_ingress_vport_metadata_update(esw, MLX5_VPORT_UPLINK,
+								 pf_metadata);
+		if (err)
+			goto err_metadata;
+	}
+
+	for (i = 0; i < ldev->ports; i++) {
+		dev = ldev->pf[i].dev;
+		mlx5_notifier_call_chain(dev->priv.events, MLX5_DEV_EVENT_MULTIPORT_ESW,
+					 (void *)0);
+	}
+
+	return 0;
+
+err_metadata:
+	mlx5_mpesw_metadata_cleanup(ldev);
+	return err;
+}
+
 static int enable_mpesw(struct mlx5_lag *ldev)
 {
 	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
@@ -21,6 +79,10 @@ static int enable_mpesw(struct mlx5_lag *ldev)
 	    !mlx5_lag_check_prereq(ldev))
 		return -EOPNOTSUPP;
 
+	err = mlx5_mpesw_metadata_set(ldev);
+	if (err)
+		return err;
+
 	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, false);
 	if (err) {
 		mlx5_core_warn(dev, "Failed to create LAG in MPESW mode (%d)\n", err);
@@ -30,13 +92,16 @@ static int enable_mpesw(struct mlx5_lag *ldev)
 	return 0;
 
 out_err:
+	mlx5_mpesw_metadata_cleanup(ldev);
 	return err;
 }
 
 static void disable_mpesw(struct mlx5_lag *ldev)
 {
-	if (ldev->mode == MLX5_LAG_MODE_MPESW)
+	if (ldev->mode == MLX5_LAG_MODE_MPESW) {
+		mlx5_mpesw_metadata_cleanup(ldev);
 		mlx5_disable_lag(ldev);
+	}
 }
 
 static void mlx5_mpesw_work(struct work_struct *work)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
index d857ea988bf2..02520f27a033 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
@@ -9,6 +9,7 @@
 
 struct lag_mpesw {
 	struct work_struct mpesw_work;
+	u32 pf_metadata[MLX5_MAX_PORTS];
 };
 
 enum mpesw_op {
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index ecd3b5448fe9..a4bb5842a948 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -217,6 +217,7 @@ struct mlx5_rsc_debug {
 enum mlx5_dev_event {
 	MLX5_DEV_EVENT_SYS_ERROR = 128, /* 0 - 127 are FW events */
 	MLX5_DEV_EVENT_PORT_AFFINITY = 129,
+	MLX5_DEV_EVENT_MULTIPORT_ESW = 130,
 };
 
 enum mlx5_port_status {
-- 
2.39.1

[net-next 05/15] net/mlx5: Lag, Add single RDMA device in multiport mode

[Saeed Mahameed]

From: Mark Bloch <mbloch@nvidia.com>

In MultiPort E-Switch mode a single RDMA is created. This device has multiple
RDMA ports that represent the uplink ports that are connected to the E-Switch.
Account for this when creating the RDMA device so it has an addtional port for
the non native uplink.

As a side effect of this patch, use shared fdb in multiport eswitch mode.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/infiniband/hw/mlx5/ib_rep.c           | 18 ++++++---
 .../net/ethernet/mellanox/mlx5/core/lag/lag.c |  6 +--
 .../net/ethernet/mellanox/mlx5/core/lag/lag.h |  3 ++
 .../ethernet/mellanox/mlx5/core/lag/mpesw.c   | 37 +++++++++++++++----
 include/linux/mlx5/driver.h                   |  1 +
 5 files changed, 49 insertions(+), 16 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/ib_rep.c b/drivers/infiniband/hw/mlx5/ib_rep.c
index 52821485371a..ddcfc116b19a 100644
--- a/drivers/infiniband/hw/mlx5/ib_rep.c
+++ b/drivers/infiniband/hw/mlx5/ib_rep.c
@@ -37,6 +37,7 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 	const struct mlx5_ib_profile *profile;
 	struct mlx5_core_dev *peer_dev;
 	struct mlx5_ib_dev *ibdev;
+	int second_uplink = false;
 	u32 peer_num_ports;
 	int vport_index;
 	int ret;
@@ -47,17 +48,24 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 		peer_dev = mlx5_lag_get_peer_mdev(dev);
 		peer_num_ports = mlx5_eswitch_get_total_vports(peer_dev);
 		if (mlx5_lag_is_master(dev)) {
-			/* Only 1 ib port is the representor for both uplinks */
-			num_ports += peer_num_ports - 1;
+			if (mlx5_lag_is_mpesw(dev))
+				num_ports += peer_num_ports;
+			else
+				num_ports += peer_num_ports - 1;
+
 		} else {
-			if (rep->vport == MLX5_VPORT_UPLINK)
-				return 0;
+			if (rep->vport == MLX5_VPORT_UPLINK) {
+				if (!mlx5_lag_is_mpesw(dev))
+					return 0;
+				second_uplink = true;
+			}
+
 			vport_index += peer_num_ports;
 			dev = peer_dev;
 		}
 	}
 
-	if (rep->vport == MLX5_VPORT_UPLINK)
+	if (rep->vport == MLX5_VPORT_UPLINK && !second_uplink)
 		profile = &raw_eth_profile;
 	else
 		return mlx5_ib_set_vport_rep(dev, rep, vport_index);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
index 301994741b08..5d331b940f4d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
@@ -644,7 +644,7 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
 	return 0;
 }
 
-static int mlx5_deactivate_lag(struct mlx5_lag *ldev)
+int mlx5_deactivate_lag(struct mlx5_lag *ldev)
 {
 	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
 	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
@@ -721,7 +721,7 @@ bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
 	return true;
 }
 
-static void mlx5_lag_add_devices(struct mlx5_lag *ldev)
+void mlx5_lag_add_devices(struct mlx5_lag *ldev)
 {
 	int i;
 
@@ -738,7 +738,7 @@ static void mlx5_lag_add_devices(struct mlx5_lag *ldev)
 	}
 }
 
-static void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
+void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
 {
 	int i;
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
index 2dbd96a86ef8..bc1f1dd3e283 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
@@ -120,5 +120,8 @@ void mlx5_infer_tx_enabled(struct lag_tracker *tracker, u8 num_ports,
 void mlx5_ldev_add_debugfs(struct mlx5_core_dev *dev);
 void mlx5_ldev_remove_debugfs(struct dentry *dbg);
 void mlx5_disable_lag(struct mlx5_lag *ldev);
+void mlx5_lag_remove_devices(struct mlx5_lag *ldev);
+int mlx5_deactivate_lag(struct mlx5_lag *ldev);
+void mlx5_lag_add_devices(struct mlx5_lag *ldev);
 
 #endif /* __MLX5_LAG_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index 2f7f2af312d7..0c0ef600f643 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -67,15 +67,16 @@ static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
 
 static int enable_mpesw(struct mlx5_lag *ldev)
 {
-	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
 	int err;
 
 	if (ldev->mode != MLX5_LAG_MODE_NONE)
 		return -EINVAL;
 
-	if (mlx5_eswitch_mode(dev) != MLX5_ESWITCH_OFFLOADS ||
-	    !MLX5_CAP_PORT_SELECTION(dev, port_select_flow_table) ||
-	    !MLX5_CAP_GEN(dev, create_lag_when_not_master_up) ||
+	if (mlx5_eswitch_mode(dev0) != MLX5_ESWITCH_OFFLOADS ||
+	    !MLX5_CAP_PORT_SELECTION(dev0, port_select_flow_table) ||
+	    !MLX5_CAP_GEN(dev0, create_lag_when_not_master_up) ||
 	    !mlx5_lag_check_prereq(ldev))
 		return -EOPNOTSUPP;
 
@@ -83,15 +84,32 @@ static int enable_mpesw(struct mlx5_lag *ldev)
 	if (err)
 		return err;
 
-	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, false);
+	mlx5_lag_remove_devices(ldev);
+
+	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, true);
 	if (err) {
-		mlx5_core_warn(dev, "Failed to create LAG in MPESW mode (%d)\n", err);
-		goto out_err;
+		mlx5_core_warn(dev0, "Failed to create LAG in MPESW mode (%d)\n", err);
+		goto err_add_devices;
 	}
 
+	dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+	mlx5_rescan_drivers_locked(dev0);
+	err = mlx5_eswitch_reload_reps(dev0->priv.eswitch);
+	if (!err)
+		err = mlx5_eswitch_reload_reps(dev1->priv.eswitch);
+	if (err)
+		goto err_rescan_drivers;
+
 	return 0;
 
-out_err:
+err_rescan_drivers:
+	dev0->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+	mlx5_rescan_drivers_locked(dev0);
+	mlx5_deactivate_lag(ldev);
+err_add_devices:
+	mlx5_lag_add_devices(ldev);
+	mlx5_eswitch_reload_reps(dev0->priv.eswitch);
+	mlx5_eswitch_reload_reps(dev1->priv.eswitch);
 	mlx5_mpesw_metadata_cleanup(ldev);
 	return err;
 }
@@ -109,6 +127,7 @@ static void mlx5_mpesw_work(struct work_struct *work)
 	struct mlx5_mpesw_work_st *mpesww = container_of(work, struct mlx5_mpesw_work_st, work);
 	struct mlx5_lag *ldev = mpesww->lag;
 
+	mlx5_dev_list_lock();
 	mutex_lock(&ldev->lock);
 	if (ldev->mode_changes_in_progress) {
 		mpesww->result = -EAGAIN;
@@ -121,6 +140,7 @@ static void mlx5_mpesw_work(struct work_struct *work)
 		disable_mpesw(ldev);
 unlock:
 	mutex_unlock(&ldev->lock);
+	mlx5_dev_list_unlock();
 	complete(&mpesww->comp);
 }
 
@@ -187,3 +207,4 @@ bool mlx5_lag_is_mpesw(struct mlx5_core_dev *dev)
 
 	return ldev && ldev->mode == MLX5_LAG_MODE_MPESW;
 }
+EXPORT_SYMBOL(mlx5_lag_is_mpesw);
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index a4bb5842a948..c9259350cdfc 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -1162,6 +1162,7 @@ bool mlx5_lag_is_active(struct mlx5_core_dev *dev);
 bool mlx5_lag_mode_is_hash(struct mlx5_core_dev *dev);
 bool mlx5_lag_is_master(struct mlx5_core_dev *dev);
 bool mlx5_lag_is_shared_fdb(struct mlx5_core_dev *dev);
+bool mlx5_lag_is_mpesw(struct mlx5_core_dev *dev);
 struct net_device *mlx5_lag_get_roce_netdev(struct mlx5_core_dev *dev);
 u8 mlx5_lag_get_slave_port(struct mlx5_core_dev *dev,
 			   struct net_device *slave);
-- 
2.39.1

[net-next 06/15] net/mlx5e: Use a simpler comparison for uplink rep

[Saeed Mahameed]

From: Roi Dayan <roid@nvidia.com>

get_route_and_out_devs()  is uses the following condition
mlx5e_eswitch_rep() && mlx5e_is_uplink_rep() to check if a given netdev is the
uplink rep.

Alternatively we can just use the straight forward version
mlx5e_eswitch_uplink_rep() that only checks if a given netdev is uplink rep.

Signed-off-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
index 684c0293a4d6..00a04fdd756f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
@@ -93,8 +93,7 @@ static int get_route_and_out_devs(struct mlx5e_priv *priv,
 	else
 		return -EOPNOTSUPP;
 
-	if (!(mlx5e_eswitch_rep(*out_dev) &&
-	      mlx5e_is_uplink_rep(netdev_priv(*out_dev))))
+	if (!mlx5e_eswitch_uplink_rep(*out_dev))
 		return -EOPNOTSUPP;
 
 	if (mlx5e_eswitch_uplink_rep(priv->netdev) && *out_dev != priv->netdev &&
-- 
2.39.1

[net-next 07/15] net/mlx5e: TC, Remove redundant parse_attr argument

[Saeed Mahameed]

From: Roi Dayan <roid@nvidia.com>

The parse_attr argument is not being used in
actions_match_supported_fdb(). remove it.

Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 780b2aa2ace1..7c714f4059c9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -3650,7 +3650,6 @@ static bool modify_header_match_supported(struct mlx5e_priv *priv,
 
 static bool
 actions_match_supported_fdb(struct mlx5e_priv *priv,
-			    struct mlx5e_tc_flow_parse_attr *parse_attr,
 			    struct mlx5e_tc_flow *flow,
 			    struct netlink_ext_ack *extack)
 {
@@ -3699,7 +3698,7 @@ actions_match_supported(struct mlx5e_priv *priv,
 		return false;
 
 	if (mlx5e_is_eswitch_flow(flow) &&
-	    !actions_match_supported_fdb(priv, parse_attr, flow, extack))
+	    !actions_match_supported_fdb(priv, flow, extack))
 		return false;
 
 	return true;
-- 
2.39.1

[net-next 08/15] net/mlx5: Remove outdated comment

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

The comment is no longer applicable, as the devlink reload and instance
cleanup are both protected with devlink instance lock, therefore no race
can happen.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/dev.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/dev.c b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
index 17ae9b4ec794..49bbfadc8c64 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/dev.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
@@ -377,10 +377,6 @@ int mlx5_attach_device(struct mlx5_core_dev *dev)
 
 			/* Pay attention that this is not PCI driver that
 			 * mlx5_core_dev is connected, but auxiliary driver.
-			 *
-			 * Here we can race of module unload with devlink
-			 * reload, but we don't need to take extra lock because
-			 * we are holding global mlx5_intf_mutex.
 			 */
 			if (!adev->dev.driver)
 				continue;
-- 
2.39.1

[net-next 09/15] net/mlx5e: Pass mdev to mlx5e_devlink_port_register()

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

Instead of accessing priv->mdev, pass mdev pointer to
mlx5e_devlink_port_register() and access it directly.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/en/devlink.c  | 15 ++++++++-------
 .../net/ethernet/mellanox/mlx5/core/en/devlink.h  |  3 ++-
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c |  2 +-
 3 files changed, 11 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
index 03ad3b61dfc7..68502c742311 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
@@ -38,7 +38,8 @@ mlx5e_devlink_get_port_parent_id(struct mlx5_core_dev *dev, struct netdev_phys_i
 }
 
 int mlx5e_devlink_port_register(struct mlx5e_dev *mlx5e_dev,
-				struct mlx5e_priv *priv)
+				struct mlx5e_priv *priv,
+				struct mlx5_core_dev *mdev)
 {
 	struct devlink *devlink = priv_to_devlink(mlx5e_dev);
 	struct devlink_port_attrs attrs = {};
@@ -46,19 +47,19 @@ int mlx5e_devlink_port_register(struct mlx5e_dev *mlx5e_dev,
 	struct devlink_port *dl_port;
 	unsigned int dl_port_index;
 
-	if (mlx5_core_is_pf(priv->mdev)) {
+	if (mlx5_core_is_pf(mdev)) {
 		attrs.flavour = DEVLINK_PORT_FLAVOUR_PHYSICAL;
-		attrs.phys.port_number = mlx5_get_dev_index(priv->mdev);
-		if (MLX5_ESWITCH_MANAGER(priv->mdev)) {
-			mlx5e_devlink_get_port_parent_id(priv->mdev, &ppid);
+		attrs.phys.port_number = mlx5_get_dev_index(mdev);
+		if (MLX5_ESWITCH_MANAGER(mdev)) {
+			mlx5e_devlink_get_port_parent_id(mdev, &ppid);
 			memcpy(attrs.switch_id.id, ppid.id, ppid.id_len);
 			attrs.switch_id.id_len = ppid.id_len;
 		}
-		dl_port_index = mlx5_esw_vport_to_devlink_port_index(priv->mdev,
+		dl_port_index = mlx5_esw_vport_to_devlink_port_index(mdev,
 								     MLX5_VPORT_UPLINK);
 	} else {
 		attrs.flavour = DEVLINK_PORT_FLAVOUR_VIRTUAL;
-		dl_port_index = mlx5_esw_vport_to_devlink_port_index(priv->mdev, 0);
+		dl_port_index = mlx5_esw_vport_to_devlink_port_index(mdev, 0);
 	}
 
 	dl_port = mlx5e_devlink_get_dl_port(priv);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
index 19b1d8e9634e..d1088c8de0d3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
@@ -10,7 +10,8 @@
 struct mlx5e_dev *mlx5e_create_devlink(struct device *dev);
 void mlx5e_destroy_devlink(struct mlx5e_dev *mlx5e_dev);
 int mlx5e_devlink_port_register(struct mlx5e_dev *mlx5e_dev,
-				struct mlx5e_priv *priv);
+				struct mlx5e_priv *priv,
+				struct mlx5_core_dev *mdev);
 void mlx5e_devlink_port_unregister(struct mlx5e_priv *priv);
 
 static inline struct devlink_port *
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 7499d64f6ec6..8c541d5055ef 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5910,7 +5910,7 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 	priv->dfs_root = debugfs_create_dir("nic",
 					    mlx5_debugfs_get_dev_root(priv->mdev));
 
-	err = mlx5e_devlink_port_register(mlx5e_dev, priv);
+	err = mlx5e_devlink_port_register(mlx5e_dev, priv, mdev);
 	if (err) {
 		mlx5_core_err(mdev, "mlx5e_devlink_port_register failed, %d\n", err);
 		goto err_destroy_netdev;
-- 
2.39.1

[net-next 10/15] net/mlx5e: Replace usage of mlx5e_devlink_get_dl_port() by netdev->devlink_port

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

On places where netdev pointer is available, access related devlink_port
pointer by netdev->devlink_port instead of using
mlx5e_devlink_get_dl_port() which is going to be removed.

Move SET_NETDEV_DEVLINK_PORT() call right after devlink port
registration to make sure netdev->devlink_port is valid.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c | 4 ++--
 drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c | 4 ++--
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c        | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_rx.c          | 6 ++----
 4 files changed, 7 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c
index 95edab4a1732..c462fe76495b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c
@@ -736,10 +736,10 @@ static const struct devlink_health_reporter_ops mlx5_rx_reporter_ops = {
 
 void mlx5e_reporter_rx_create(struct mlx5e_priv *priv)
 {
-	struct devlink_port *dl_port = mlx5e_devlink_get_dl_port(priv);
 	struct devlink_health_reporter *reporter;
 
-	reporter = devlink_port_health_reporter_create(dl_port, &mlx5_rx_reporter_ops,
+	reporter = devlink_port_health_reporter_create(priv->netdev->devlink_port,
+						       &mlx5_rx_reporter_ops,
 						       MLX5E_REPORTER_RX_GRACEFUL_PERIOD, priv);
 	if (IS_ERR(reporter)) {
 		netdev_warn(priv->netdev, "Failed to create rx reporter, err = %ld\n",
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
index 41e356d9d785..34666e2b3871 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
@@ -594,10 +594,10 @@ static const struct devlink_health_reporter_ops mlx5_tx_reporter_ops = {
 
 void mlx5e_reporter_tx_create(struct mlx5e_priv *priv)
 {
-	struct devlink_port *dl_port = mlx5e_devlink_get_dl_port(priv);
 	struct devlink_health_reporter *reporter;
 
-	reporter = devlink_port_health_reporter_create(dl_port, &mlx5_tx_reporter_ops,
+	reporter = devlink_port_health_reporter_create(priv->netdev->devlink_port,
+						       &mlx5_tx_reporter_ops,
 						       MLX5_REPORTER_TX_GRACEFUL_PERIOD, priv);
 	if (IS_ERR(reporter)) {
 		netdev_warn(priv->netdev,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 8c541d5055ef..18b2fe658d20 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5915,6 +5915,7 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 		mlx5_core_err(mdev, "mlx5e_devlink_port_register failed, %d\n", err);
 		goto err_destroy_netdev;
 	}
+	SET_NETDEV_DEVLINK_PORT(netdev, mlx5e_devlink_get_dl_port(priv));
 
 	err = profile->init(mdev, netdev);
 	if (err) {
@@ -5928,7 +5929,6 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 		goto err_profile_cleanup;
 	}
 
-	SET_NETDEV_DEVLINK_PORT(netdev, mlx5e_devlink_get_dl_port(priv));
 	err = register_netdev(netdev);
 	if (err) {
 		mlx5_core_err(mdev, "register_netdev failed, %d\n", err);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index a9473a51edc1..b2c7ec4692f0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -2570,10 +2570,8 @@ int mlx5e_rq_set_handlers(struct mlx5e_rq *rq, struct mlx5e_params *params, bool
 
 static void mlx5e_trap_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
 {
-	struct mlx5e_priv *priv = netdev_priv(rq->netdev);
 	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
 	struct mlx5e_wqe_frag_info *wi;
-	struct devlink_port *dl_port;
 	struct sk_buff *skb;
 	u32 cqe_bcnt;
 	u16 trap_id;
@@ -2596,8 +2594,8 @@ static void mlx5e_trap_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe
 	mlx5e_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
 	skb_push(skb, ETH_HLEN);
 
-	dl_port = mlx5e_devlink_get_dl_port(priv);
-	mlx5_devlink_trap_report(rq->mdev, trap_id, skb, dl_port);
+	mlx5_devlink_trap_report(rq->mdev, trap_id, skb,
+				 rq->netdev->devlink_port);
 	dev_kfree_skb_any(skb);
 
 free_wqe:
-- 
2.39.1

[net-next 11/15] net/mlx5e: Move dl_port to struct mlx5e_dev

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

No need to have dl_port which is tightly coupled with mlx5e code
in mlx5 core code. Move it to struct mlx5e_dev and loose
mlx5e_devlink_get_dl_port() helper.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en.h      |  1 +
 .../net/ethernet/mellanox/mlx5/core/en/devlink.c  | 15 +++++----------
 .../net/ethernet/mellanox/mlx5/core/en/devlink.h  |  9 +--------
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c |  8 ++++----
 include/linux/mlx5/driver.h                       |  1 -
 5 files changed, 11 insertions(+), 23 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 125c7cb7d839..88460b7796e5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -975,6 +975,7 @@ struct mlx5e_priv {
 
 struct mlx5e_dev {
 	struct mlx5e_priv *priv;
+	struct devlink_port dl_port;
 };
 
 struct mlx5e_rx_handlers {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
index 68502c742311..724de9e06c54 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
@@ -38,13 +38,11 @@ mlx5e_devlink_get_port_parent_id(struct mlx5_core_dev *dev, struct netdev_phys_i
 }
 
 int mlx5e_devlink_port_register(struct mlx5e_dev *mlx5e_dev,
-				struct mlx5e_priv *priv,
 				struct mlx5_core_dev *mdev)
 {
 	struct devlink *devlink = priv_to_devlink(mlx5e_dev);
 	struct devlink_port_attrs attrs = {};
 	struct netdev_phys_item_id ppid = {};
-	struct devlink_port *dl_port;
 	unsigned int dl_port_index;
 
 	if (mlx5_core_is_pf(mdev)) {
@@ -62,16 +60,13 @@ int mlx5e_devlink_port_register(struct mlx5e_dev *mlx5e_dev,
 		dl_port_index = mlx5_esw_vport_to_devlink_port_index(mdev, 0);
 	}
 
-	dl_port = mlx5e_devlink_get_dl_port(priv);
-	memset(dl_port, 0, sizeof(*dl_port));
-	devlink_port_attrs_set(dl_port, &attrs);
+	devlink_port_attrs_set(&mlx5e_dev->dl_port, &attrs);
 
-	return devlink_port_register(devlink, dl_port, dl_port_index);
+	return devlink_port_register(devlink, &mlx5e_dev->dl_port,
+				     dl_port_index);
 }
 
-void mlx5e_devlink_port_unregister(struct mlx5e_priv *priv)
+void mlx5e_devlink_port_unregister(struct mlx5e_dev *mlx5e_dev)
 {
-	struct devlink_port *dl_port = mlx5e_devlink_get_dl_port(priv);
-
-	devlink_port_unregister(dl_port);
+	devlink_port_unregister(&mlx5e_dev->dl_port);
 }
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
index d1088c8de0d3..c31d1d97b8c7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
@@ -10,14 +10,7 @@
 struct mlx5e_dev *mlx5e_create_devlink(struct device *dev);
 void mlx5e_destroy_devlink(struct mlx5e_dev *mlx5e_dev);
 int mlx5e_devlink_port_register(struct mlx5e_dev *mlx5e_dev,
-				struct mlx5e_priv *priv,
 				struct mlx5_core_dev *mdev);
-void mlx5e_devlink_port_unregister(struct mlx5e_priv *priv);
-
-static inline struct devlink_port *
-mlx5e_devlink_get_dl_port(struct mlx5e_priv *priv)
-{
-	return &priv->mdev->mlx5e_res.dl_port;
-}
+void mlx5e_devlink_port_unregister(struct mlx5e_dev *mlx5e_dev);
 
 #endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 18b2fe658d20..432d32f5dc65 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5910,12 +5910,12 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 	priv->dfs_root = debugfs_create_dir("nic",
 					    mlx5_debugfs_get_dev_root(priv->mdev));
 
-	err = mlx5e_devlink_port_register(mlx5e_dev, priv, mdev);
+	err = mlx5e_devlink_port_register(mlx5e_dev, mdev);
 	if (err) {
 		mlx5_core_err(mdev, "mlx5e_devlink_port_register failed, %d\n", err);
 		goto err_destroy_netdev;
 	}
-	SET_NETDEV_DEVLINK_PORT(netdev, mlx5e_devlink_get_dl_port(priv));
+	SET_NETDEV_DEVLINK_PORT(netdev, &mlx5e_dev->dl_port);
 
 	err = profile->init(mdev, netdev);
 	if (err) {
@@ -5945,7 +5945,7 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 err_profile_cleanup:
 	profile->cleanup(priv);
 err_devlink_cleanup:
-	mlx5e_devlink_port_unregister(priv);
+	mlx5e_devlink_port_unregister(mlx5e_dev);
 err_destroy_netdev:
 	debugfs_remove_recursive(priv->dfs_root);
 	mlx5e_destroy_netdev(priv);
@@ -5965,7 +5965,7 @@ static void mlx5e_remove(struct auxiliary_device *adev)
 	unregister_netdev(priv->netdev);
 	mlx5e_suspend(adev, state);
 	priv->profile->cleanup(priv);
-	mlx5e_devlink_port_unregister(priv);
+	mlx5e_devlink_port_unregister(mlx5e_dev);
 	debugfs_remove_recursive(priv->dfs_root);
 	mlx5e_destroy_netdev(priv);
 	mlx5e_destroy_devlink(mlx5e_dev);
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index c9259350cdfc..a170c8565779 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -679,7 +679,6 @@ struct mlx5e_resources {
 		u32			   mkey;
 		struct mlx5_sq_bfreg       bfreg;
 	} hw_objs;
-	struct devlink_port dl_port;
 	struct net_device *uplink_netdev;
 	struct mutex uplink_netdev_lock;
 	struct mlx5_crypto_dek_priv *dek_priv;
-- 
2.39.1

[net-next 12/15] net/mlx5e: Move devlink port registration to be done before netdev alloc

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

Move the devlink port registration to be done right after devlink
instance registration.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_main.c | 24 +++++++++----------
 1 file changed, 12 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 432d32f5dc65..9657c66c8f69 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5892,12 +5892,19 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 		return PTR_ERR(mlx5e_dev);
 	auxiliary_set_drvdata(adev, mlx5e_dev);
 
+	err = mlx5e_devlink_port_register(mlx5e_dev, mdev);
+	if (err) {
+		mlx5_core_err(mdev, "mlx5e_devlink_port_register failed, %d\n", err);
+		goto err_devlink_unregister;
+	}
+
 	netdev = mlx5e_create_netdev(mdev, profile);
 	if (!netdev) {
 		mlx5_core_err(mdev, "mlx5e_create_netdev failed\n");
 		err = -ENOMEM;
-		goto err_devlink_unregister;
+		goto err_devlink_port_unregister;
 	}
+	SET_NETDEV_DEVLINK_PORT(netdev, &mlx5e_dev->dl_port);
 
 	mlx5e_build_nic_netdev(netdev);
 
@@ -5910,17 +5917,10 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 	priv->dfs_root = debugfs_create_dir("nic",
 					    mlx5_debugfs_get_dev_root(priv->mdev));
 
-	err = mlx5e_devlink_port_register(mlx5e_dev, mdev);
-	if (err) {
-		mlx5_core_err(mdev, "mlx5e_devlink_port_register failed, %d\n", err);
-		goto err_destroy_netdev;
-	}
-	SET_NETDEV_DEVLINK_PORT(netdev, &mlx5e_dev->dl_port);
-
 	err = profile->init(mdev, netdev);
 	if (err) {
 		mlx5_core_err(mdev, "mlx5e_nic_profile init failed, %d\n", err);
-		goto err_devlink_cleanup;
+		goto err_destroy_netdev;
 	}
 
 	err = mlx5e_resume(adev);
@@ -5944,11 +5944,11 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 	mlx5e_suspend(adev, state);
 err_profile_cleanup:
 	profile->cleanup(priv);
-err_devlink_cleanup:
-	mlx5e_devlink_port_unregister(mlx5e_dev);
 err_destroy_netdev:
 	debugfs_remove_recursive(priv->dfs_root);
 	mlx5e_destroy_netdev(priv);
+err_devlink_port_unregister:
+	mlx5e_devlink_port_unregister(mlx5e_dev);
 err_devlink_unregister:
 	mlx5e_destroy_devlink(mlx5e_dev);
 	return err;
@@ -5965,9 +5965,9 @@ static void mlx5e_remove(struct auxiliary_device *adev)
 	unregister_netdev(priv->netdev);
 	mlx5e_suspend(adev, state);
 	priv->profile->cleanup(priv);
-	mlx5e_devlink_port_unregister(mlx5e_dev);
 	debugfs_remove_recursive(priv->dfs_root);
 	mlx5e_destroy_netdev(priv);
+	mlx5e_devlink_port_unregister(mlx5e_dev);
 	mlx5e_destroy_devlink(mlx5e_dev);
 }
 
-- 
2.39.1

[net-next 13/15] net/mlx5e: Create auxdev devlink instance in the same ns as parent devlink

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

Commit cited in "fixes" tag moved the devlink port under separate
devlink entity created for auxiliary device. Respect the network
namespace of parent devlink entity and allocate the devlink there.

Fixes: ee75f1fc44dd ("net/mlx5e: Create separate devlink instance for ethernet auxiliary device")
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c | 6 ++++--
 drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h | 3 ++-
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c    | 2 +-
 3 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
index 724de9e06c54..c6b6e290fd79 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
@@ -7,12 +7,14 @@
 static const struct devlink_ops mlx5e_devlink_ops = {
 };
 
-struct mlx5e_dev *mlx5e_create_devlink(struct device *dev)
+struct mlx5e_dev *mlx5e_create_devlink(struct device *dev,
+				       struct mlx5_core_dev *mdev)
 {
 	struct mlx5e_dev *mlx5e_dev;
 	struct devlink *devlink;
 
-	devlink = devlink_alloc(&mlx5e_devlink_ops, sizeof(*mlx5e_dev), dev);
+	devlink = devlink_alloc_ns(&mlx5e_devlink_ops, sizeof(*mlx5e_dev),
+				   devlink_net(priv_to_devlink(mdev)), dev);
 	if (!devlink)
 		return ERR_PTR(-ENOMEM);
 	devlink_register(devlink);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
index c31d1d97b8c7..d5ec4461f300 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
@@ -7,7 +7,8 @@
 #include <net/devlink.h>
 #include "en.h"
 
-struct mlx5e_dev *mlx5e_create_devlink(struct device *dev);
+struct mlx5e_dev *mlx5e_create_devlink(struct device *dev,
+				       struct mlx5_core_dev *mdev);
 void mlx5e_destroy_devlink(struct mlx5e_dev *mlx5e_dev);
 int mlx5e_devlink_port_register(struct mlx5e_dev *mlx5e_dev,
 				struct mlx5_core_dev *mdev);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 9657c66c8f69..28120c4a9746 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5887,7 +5887,7 @@ static int mlx5e_probe(struct auxiliary_device *adev,
 	struct mlx5e_priv *priv;
 	int err;
 
-	mlx5e_dev = mlx5e_create_devlink(&adev->dev);
+	mlx5e_dev = mlx5e_create_devlink(&adev->dev, mdev);
 	if (IS_ERR(mlx5e_dev))
 		return PTR_ERR(mlx5e_dev);
 	auxiliary_set_drvdata(adev, mlx5e_dev);
-- 
2.39.1

[net-next 14/15] net/mlx5: Remove "recovery" arg from mlx5_load_one() function

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

mlx5_load_one() is always called with recovery==false, so remove the
unneeded function arg.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c  | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/main.c      | 9 ++++-----
 drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h | 2 +-
 3 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
index 63290da84010..7bde34a17165 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
@@ -167,7 +167,7 @@ static void mlx5_fw_reset_complete_reload(struct mlx5_core_dev *dev)
 		if (mlx5_health_wait_pci_up(dev))
 			mlx5_core_err(dev, "reset reload flow aborted, PCI reads still not working\n");
 		else
-			mlx5_load_one(dev, false);
+			mlx5_load_one(dev);
 		devlink_remote_reload_actions_performed(priv_to_devlink(dev), 0,
 							BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT) |
 							BIT(DEVLINK_RELOAD_ACTION_FW_ACTIVATE));
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index 26e1057845fe..6997b29fdecc 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -1509,13 +1509,13 @@ int mlx5_load_one_devl_locked(struct mlx5_core_dev *dev, bool recovery)
 	return err;
 }
 
-int mlx5_load_one(struct mlx5_core_dev *dev, bool recovery)
+int mlx5_load_one(struct mlx5_core_dev *dev)
 {
 	struct devlink *devlink = priv_to_devlink(dev);
 	int ret;
 
 	devl_lock(devlink);
-	ret = mlx5_load_one_devl_locked(dev, recovery);
+	ret = mlx5_load_one_devl_locked(dev, false);
 	devl_unlock(devlink);
 	return ret;
 }
@@ -1912,8 +1912,7 @@ static void mlx5_pci_resume(struct pci_dev *pdev)
 
 	mlx5_pci_trace(dev, "Enter, loading driver..\n");
 
-	err = mlx5_load_one(dev, false);
-
+	err = mlx5_load_one(dev);
 	if (!err)
 		devlink_health_reporter_state_update(dev->priv.health.fw_fatal_reporter,
 						     DEVLINK_HEALTH_REPORTER_STATE_HEALTHY);
@@ -2004,7 +2003,7 @@ static int mlx5_resume(struct pci_dev *pdev)
 {
 	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
 
-	return mlx5_load_one(dev, false);
+	return mlx5_load_one(dev);
 }
 
 static const struct pci_device_id mlx5_core_pci_table[] = {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
index 029305a8b80a..d670b28a7323 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
@@ -321,7 +321,7 @@ int mlx5_init_one(struct mlx5_core_dev *dev);
 void mlx5_uninit_one(struct mlx5_core_dev *dev);
 void mlx5_unload_one(struct mlx5_core_dev *dev);
 void mlx5_unload_one_devl_locked(struct mlx5_core_dev *dev);
-int mlx5_load_one(struct mlx5_core_dev *dev, bool recovery);
+int mlx5_load_one(struct mlx5_core_dev *dev);
 int mlx5_load_one_devl_locked(struct mlx5_core_dev *dev, bool recovery);
 
 int mlx5_vport_set_other_func_cap(struct mlx5_core_dev *dev, const void *hca_cap, u16 function_id,
-- 
2.39.1

[net-next 15/15] net/mlx5: Suspend auxiliary devices only in case of PCI device suspend

[Saeed Mahameed]

From: Jiri Pirko <jiri@nvidia.com>

The original behavior introduced by commit c6acd629eec7 ("net/mlx5e: Add
support for devlink-port in non-representors mode") correctly
re-instantiated uplink devlink port and related netdevice during devlink
reload. However with migration to auxiliary devices, this behaviour
changed.

Restore the original behaviour and tear down auxiliary devices
completely during devlink reload.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/dev.c    |  4 ++--
 .../net/ethernet/mellanox/mlx5/core/devlink.c    |  4 ++--
 .../net/ethernet/mellanox/mlx5/core/fw_reset.c   |  4 ++--
 drivers/net/ethernet/mellanox/mlx5/core/health.c |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/main.c   | 16 ++++++++--------
 .../net/ethernet/mellanox/mlx5/core/mlx5_core.h  |  6 +++---
 .../ethernet/mellanox/mlx5/core/sf/dev/driver.c  |  2 +-
 7 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/dev.c b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
index 49bbfadc8c64..445fe30c3d0b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/dev.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
@@ -396,7 +396,7 @@ int mlx5_attach_device(struct mlx5_core_dev *dev)
 	return ret;
 }
 
-void mlx5_detach_device(struct mlx5_core_dev *dev)
+void mlx5_detach_device(struct mlx5_core_dev *dev, bool suspend)
 {
 	struct mlx5_priv *priv = &dev->priv;
 	struct auxiliary_device *adev;
@@ -425,7 +425,7 @@ void mlx5_detach_device(struct mlx5_core_dev *dev)
 
 		adrv = to_auxiliary_drv(adev->dev.driver);
 
-		if (adrv->suspend) {
+		if (adrv->suspend && suspend) {
 			adrv->suspend(adev, pm);
 			continue;
 		}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
index 1bccc6c31460..7f8d518f7ced 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
@@ -105,7 +105,7 @@ static int mlx5_devlink_reload_fw_activate(struct devlink *devlink, struct netli
 	if (err)
 		return err;
 
-	mlx5_unload_one_devl_locked(dev);
+	mlx5_unload_one_devl_locked(dev, true);
 	err = mlx5_health_wait_pci_up(dev);
 	if (err)
 		NL_SET_ERR_MSG_MOD(extack, "FW activate aborted, PCI reads fail after reset");
@@ -168,7 +168,7 @@ static int mlx5_devlink_reload_down(struct devlink *devlink, bool netns_change,
 
 	switch (action) {
 	case DEVLINK_RELOAD_ACTION_DRIVER_REINIT:
-		mlx5_unload_one_devl_locked(dev);
+		mlx5_unload_one_devl_locked(dev, false);
 		break;
 	case DEVLINK_RELOAD_ACTION_FW_ACTIVATE:
 		if (limit == DEVLINK_RELOAD_LIMIT_NO_RESET)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
index 7bde34a17165..4c2dad9d7cfb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
@@ -163,7 +163,7 @@ static void mlx5_fw_reset_complete_reload(struct mlx5_core_dev *dev)
 	if (test_bit(MLX5_FW_RESET_FLAGS_PENDING_COMP, &fw_reset->reset_flags)) {
 		complete(&fw_reset->done);
 	} else {
-		mlx5_unload_one(dev);
+		mlx5_unload_one(dev, false);
 		if (mlx5_health_wait_pci_up(dev))
 			mlx5_core_err(dev, "reset reload flow aborted, PCI reads still not working\n");
 		else
@@ -498,7 +498,7 @@ int mlx5_fw_reset_wait_reset_done(struct mlx5_core_dev *dev)
 	}
 	err = fw_reset->ret;
 	if (test_and_clear_bit(MLX5_FW_RESET_FLAGS_RELOAD_REQUIRED, &fw_reset->reset_flags)) {
-		mlx5_unload_one_devl_locked(dev);
+		mlx5_unload_one_devl_locked(dev, false);
 		mlx5_load_one_devl_locked(dev, false);
 	}
 out:
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c
index 1e8bee906c31..f9438d4e43ca 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/health.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c
@@ -699,7 +699,7 @@ static void mlx5_fw_fatal_reporter_err_work(struct work_struct *work)
 		 * requests from the kernel.
 		 */
 		mlx5_core_err(dev, "Driver is in error state. Unloading\n");
-		mlx5_unload_one(dev);
+		mlx5_unload_one(dev, false);
 	}
 }
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index 6997b29fdecc..540840e80493 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -1520,12 +1520,12 @@ int mlx5_load_one(struct mlx5_core_dev *dev)
 	return ret;
 }
 
-void mlx5_unload_one_devl_locked(struct mlx5_core_dev *dev)
+void mlx5_unload_one_devl_locked(struct mlx5_core_dev *dev, bool suspend)
 {
 	devl_assert_locked(priv_to_devlink(dev));
 	mutex_lock(&dev->intf_state_mutex);
 
-	mlx5_detach_device(dev);
+	mlx5_detach_device(dev, suspend);
 
 	if (!test_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state)) {
 		mlx5_core_warn(dev, "%s: interface is down, NOP\n",
@@ -1540,12 +1540,12 @@ void mlx5_unload_one_devl_locked(struct mlx5_core_dev *dev)
 	mutex_unlock(&dev->intf_state_mutex);
 }
 
-void mlx5_unload_one(struct mlx5_core_dev *dev)
+void mlx5_unload_one(struct mlx5_core_dev *dev, bool suspend)
 {
 	struct devlink *devlink = priv_to_devlink(dev);
 
 	devl_lock(devlink);
-	mlx5_unload_one_devl_locked(dev);
+	mlx5_unload_one_devl_locked(dev, suspend);
 	devl_unlock(devlink);
 }
 
@@ -1830,7 +1830,7 @@ static pci_ers_result_t mlx5_pci_err_detected(struct pci_dev *pdev,
 
 	mlx5_enter_error_state(dev, false);
 	mlx5_error_sw_reset(dev);
-	mlx5_unload_one(dev);
+	mlx5_unload_one(dev, true);
 	mlx5_drain_health_wq(dev);
 	mlx5_pci_disable_device(dev);
 
@@ -1986,7 +1986,7 @@ static void shutdown(struct pci_dev *pdev)
 	set_bit(MLX5_BREAK_FW_WAIT, &dev->intf_state);
 	err = mlx5_try_fast_unload(dev);
 	if (err)
-		mlx5_unload_one(dev);
+		mlx5_unload_one(dev, false);
 	mlx5_pci_disable_device(dev);
 }
 
@@ -1994,7 +1994,7 @@ static int mlx5_suspend(struct pci_dev *pdev, pm_message_t state)
 {
 	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
 
-	mlx5_unload_one(dev);
+	mlx5_unload_one(dev, true);
 
 	return 0;
 }
@@ -2037,7 +2037,7 @@ MODULE_DEVICE_TABLE(pci, mlx5_core_pci_table);
 void mlx5_disable_device(struct mlx5_core_dev *dev)
 {
 	mlx5_error_sw_reset(dev);
-	mlx5_unload_one_devl_locked(dev);
+	mlx5_unload_one_devl_locked(dev, false);
 }
 
 int mlx5_recover_device(struct mlx5_core_dev *dev)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
index d670b28a7323..be0785f83083 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
@@ -236,7 +236,7 @@ void mlx5_adev_cleanup(struct mlx5_core_dev *dev);
 int mlx5_adev_init(struct mlx5_core_dev *dev);
 
 int mlx5_attach_device(struct mlx5_core_dev *dev);
-void mlx5_detach_device(struct mlx5_core_dev *dev);
+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);
 struct mlx5_core_dev *mlx5_get_next_phys_dev_lag(struct mlx5_core_dev *dev);
@@ -319,8 +319,8 @@ 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);
 void mlx5_uninit_one(struct mlx5_core_dev *dev);
-void mlx5_unload_one(struct mlx5_core_dev *dev);
-void mlx5_unload_one_devl_locked(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);
 int mlx5_load_one_devl_locked(struct mlx5_core_dev *dev, bool recovery);
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
index 7b4783ce213e..a7377619ba6f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
@@ -74,7 +74,7 @@ static void mlx5_sf_dev_shutdown(struct auxiliary_device *adev)
 {
 	struct mlx5_sf_dev *sf_dev = container_of(adev, struct mlx5_sf_dev, adev);
 
-	mlx5_unload_one(sf_dev->mdev);
+	mlx5_unload_one(sf_dev->mdev, false);
 }
 
 static const struct auxiliary_device_id mlx5_sf_dev_id_table[] = {
-- 
2.39.1

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Jakub Kicinski]

On Fri, 10 Feb 2023 14:18:07 -0800 Saeed Mahameed wrote:
> From: Roi Dayan <roid@nvidia.com>
> 
> Instead of activating multiport eswitch dynamically through
> adding a TC rule and meeting certain conditions, allow the user
> to activate it through devlink.
> This will remove the forced requirement of using TC.
> e.g. Bridge offload.
> 
> Example:
>     $ devlink dev param set pci/0000:00:0b.0 name esw_multiport value 1 \
>                   cmode runtime

The PR message has way better description of what's going on here.

> +   * - ``esw_multiport``
> +     - Boolean
> +     - runtime
> +     - Set the E-Switch lag mode to multiport.

This is just spelling out the name, not real documentation :(

IIUC the new mode is how devices _should_ work in switchdev mode,
so why not make this the default already? What's the cut-off point?

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Jiri Pirko]

Sat, Feb 11, 2023 at 05:03:29AM CET, kuba@kernel.org wrote:
>On Fri, 10 Feb 2023 14:18:07 -0800 Saeed Mahameed wrote:
>> From: Roi Dayan <roid@nvidia.com>
>> 
>> Instead of activating multiport eswitch dynamically through
>> adding a TC rule and meeting certain conditions, allow the user
>> to activate it through devlink.
>> This will remove the forced requirement of using TC.
>> e.g. Bridge offload.
>> 
>> Example:
>>     $ devlink dev param set pci/0000:00:0b.0 name esw_multiport value 1 \
>>                   cmode runtime
>
>The PR message has way better description of what's going on here.
>
>> +   * - ``esw_multiport``
>> +     - Boolean
>> +     - runtime
>> +     - Set the E-Switch lag mode to multiport.
>
>This is just spelling out the name, not real documentation :(
>
>IIUC the new mode is how devices _should_ work in switchdev mode,

Yep, I feel exactly the same.


>so why not make this the default already? What's the cut-off point?

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Mark Bloch]

On 11/02/2023 6:03, Jakub Kicinski wrote:
> On Fri, 10 Feb 2023 14:18:07 -0800 Saeed Mahameed wrote:
>> From: Roi Dayan <roid@nvidia.com>
>>
>> Instead of activating multiport eswitch dynamically through
>> adding a TC rule and meeting certain conditions, allow the user
>> to activate it through devlink.
>> This will remove the forced requirement of using TC.
>> e.g. Bridge offload.
>>
>> Example:
>>     $ devlink dev param set pci/0000:00:0b.0 name esw_multiport value 1 \
>>                   cmode runtime
> 
> The PR message has way better description of what's going on here.
> 
>> +   * - ``esw_multiport``
>> +     - Boolean
>> +     - runtime
>> +     - Set the E-Switch lag mode to multiport.
> 
> This is just spelling out the name, not real documentation :(
> 
> IIUC the new mode is how devices _should_ work in switchdev mode,
> so why not make this the default already? What's the cut-off point?

Hi Jakub,

I agree with you this definitely should be the default. That was
the plan in the beginning. Testing uncovered that making it the default
breaks users. It changes the look and feel of the driver when in switchdev
mode, the customers we've talked with are very afraid
it will break their software and actually, we've seen real breakages
and I fully expect more to pop up once this feature goes live.

We've started reaching out to customers and working with them on updating
their software but such a change takes time and honestly, we would like to
push this change out as soon as possible and start building
on top of this new mode. Once more features that are only possible in this
new mode are added it will be an even bigger incentive to move to it.

We believe this parameter will allow customers to transition to the new
mode faster as we know this is a big change, let's start the transition
as soon as possible as we know delaying it will only make things worse.
Add a flag so we can control it and in the future, once all the software
is updated switch the flag to be the default and keep it for legacy
software where updating the logic isn't possible.

I've been dealing with LAG / E-Switch code for the past few years, which is
a very big pain for me. I would like to move forward and have a software
model that makes sense and simplifies the logic inside mlx5 and this is the
best way we have right now.

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Jakub Kicinski]

On Mon, 13 Feb 2023 21:00:16 +0200 Mark Bloch wrote:
> I agree with you this definitely should be the default. That was
> the plan in the beginning. Testing uncovered that making it the default
> breaks users. It changes the look and feel of the driver when in switchdev
> mode, the customers we've talked with are very afraid
> it will break their software and actually, we've seen real breakages
> and I fully expect more to pop up once this feature goes live.

Real breakages as in bugs that are subsequently addressed or inherent
differences which customers may need to adjust their code for?
Either way we need the expectation captured in the docs - 
an "experimental" warning or examples of cases which behave differently.

> We've started reaching out to customers and working with them on updating
> their software but such a change takes time and honestly, we would like to
> push this change out as soon as possible and start building
> on top of this new mode. Once more features that are only possible in this
> new mode are added it will be an even bigger incentive to move to it.
> 
> We believe this parameter will allow customers to transition to the new
> mode faster as we know this is a big change, let's start the transition
> as soon as possible as we know delaying it will only make things worse.
> Add a flag so we can control it and in the future, once all the software
> is updated switch the flag to be the default and keep it for legacy
> software where updating the logic isn't possible.

Oh, the "legacy software where updating the logic isn't possible"
sounds concerning. Do we know what those cases are? Can we perhaps
constrain them to run on a specific version of HW (say starting with
CX8 the param will no longer be there and only the right behavior will
be supported upstream)?

I'm speaking under assumption that the document+deprecate plan is okay
with Jiri.

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Mark Bloch]

On 14/02/2023 4:02, Jakub Kicinski wrote:
> On Mon, 13 Feb 2023 21:00:16 +0200 Mark Bloch wrote:
>> I agree with you this definitely should be the default. That was
>> the plan in the beginning. Testing uncovered that making it the default
>> breaks users. It changes the look and feel of the driver when in switchdev
>> mode, the customers we've talked with are very afraid
>> it will break their software and actually, we've seen real breakages
>> and I fully expect more to pop up once this feature goes live.
> 
> Real breakages as in bugs that are subsequently addressed or inherent
> differences which customers may need to adjust their code for?

It will require changes to the code to deal with.

> Either way we need the expectation captured in the docs - 
> an "experimental" warning or examples of cases which behave differently.

Will talk with Roi and ask him to add proper documentation for this feature.

> 
>> We've started reaching out to customers and working with them on updating
>> their software but such a change takes time and honestly, we would like to
>> push this change out as soon as possible and start building
>> on top of this new mode. Once more features that are only possible in this
>> new mode are added it will be an even bigger incentive to move to it.
>>
>> We believe this parameter will allow customers to transition to the new
>> mode faster as we know this is a big change, let's start the transition
>> as soon as possible as we know delaying it will only make things worse.
>> Add a flag so we can control it and in the future, once all the software
>> is updated switch the flag to be the default and keep it for legacy
>> software where updating the logic isn't possible.
> 
> Oh, the "legacy software where updating the logic isn't possible"
> sounds concerning. Do we know what those cases are? Can we perhaps
> constrain them to run on a specific version of HW (say starting with
> CX8 the param will no longer be there and only the right behavior will
> be supported upstream)?

While I can encourage customers to update their software to deal with
this new mode it's up to them. As you already know, it's hard to change
customers infrastructure so quickly. These things take time. They plan
on running multiple hardware generations on top of the same software.
I'll keep pushing on making this mode the default one.

> 
> I'm speaking under assumption that the document+deprecate plan is okay
> with Jiri.

Will talk with Jiri and get his input.

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Jiri Pirko]

Tue, Feb 14, 2023 at 03:02:46AM CET, kuba@kernel.org wrote:
>On Mon, 13 Feb 2023 21:00:16 +0200 Mark Bloch wrote:
>> I agree with you this definitely should be the default. That was
>> the plan in the beginning. Testing uncovered that making it the default
>> breaks users. It changes the look and feel of the driver when in switchdev
>> mode, the customers we've talked with are very afraid
>> it will break their software and actually, we've seen real breakages
>> and I fully expect more to pop up once this feature goes live.
>
>Real breakages as in bugs that are subsequently addressed or inherent
>differences which customers may need to adjust their code for?
>Either way we need the expectation captured in the docs - 
>an "experimental" warning or examples of cases which behave differently.
>
>> We've started reaching out to customers and working with them on updating
>> their software but such a change takes time and honestly, we would like to
>> push this change out as soon as possible and start building
>> on top of this new mode. Once more features that are only possible in this
>> new mode are added it will be an even bigger incentive to move to it.
>> 
>> We believe this parameter will allow customers to transition to the new
>> mode faster as we know this is a big change, let's start the transition
>> as soon as possible as we know delaying it will only make things worse.
>> Add a flag so we can control it and in the future, once all the software
>> is updated switch the flag to be the default and keep it for legacy
>> software where updating the logic isn't possible.
>
>Oh, the "legacy software where updating the logic isn't possible"
>sounds concerning. Do we know what those cases are? Can we perhaps
>constrain them to run on a specific version of HW (say starting with
>CX8 the param will no longer be there and only the right behavior will
>be supported upstream)?
>
>I'm speaking under assumption that the document+deprecate plan is okay
>with Jiri.

I just talked with Mark. Makes sense to properly document and deprecate
as soon as possible. I don't see any other way. I can't wait when it is
done :)

so +1

Thanks!

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Alexander Lobakin]

From: Saeed Mahameed <saeed@kernel.org>
Date: Fri, 10 Feb 2023 14:18:07 -0800

> From: Roi Dayan <roid@nvidia.com>
> 
> Instead of activating multiport eswitch dynamically through
> adding a TC rule and meeting certain conditions, allow the user
> to activate it through devlink.
> This will remove the forced requirement of using TC.
> e.g. Bridge offload.
> 
> Example:
>     $ devlink dev param set pci/0000:00:0b.0 name esw_multiport value 1 \
>                   cmode runtime
> 
> Signed-off-by: Roi Dayan <roid@nvidia.com>
> Reviewed-by: Maor Dickman <maord@nvidia.com>
> Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
> ---
>  Documentation/networking/devlink/mlx5.rst     |  4 ++
>  .../net/ethernet/mellanox/mlx5/core/devlink.c | 56 +++++++++++++++++++
>  .../net/ethernet/mellanox/mlx5/core/devlink.h |  1 +
>  .../mellanox/mlx5/core/en/tc/act/mirred.c     |  9 ---
>  .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 22 +-------
>  .../net/ethernet/mellanox/mlx5/core/en_tc.h   |  6 --
>  .../net/ethernet/mellanox/mlx5/core/lag/lag.c |  4 +-
>  .../net/ethernet/mellanox/mlx5/core/lag/lag.h |  1 +
>  .../ethernet/mellanox/mlx5/core/lag/mpesw.c   | 46 +++++++--------
>  .../ethernet/mellanox/mlx5/core/lag/mpesw.h   | 12 +---
>  10 files changed, 87 insertions(+), 74 deletions(-)
> 
> diff --git a/Documentation/networking/devlink/mlx5.rst b/Documentation/networking/devlink/mlx5.rst
> index 29ad304e6fba..1d2ad2727da1 100644
> --- a/Documentation/networking/devlink/mlx5.rst
> +++ b/Documentation/networking/devlink/mlx5.rst
> @@ -54,6 +54,10 @@ parameters.
>       - Control the number of large groups (size > 1) in the FDB table.
>  
>         * The default value is 15, and the range is between 1 and 1024.
> +   * - ``esw_multiport``
> +     - Boolean
> +     - runtime
> +     - Set the E-Switch lag mode to multiport.
>  
>  The ``mlx5`` driver supports reloading via ``DEVLINK_CMD_RELOAD``
>  
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
> index b742e04deec1..49392870f695 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
> @@ -7,6 +7,7 @@
>  #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"
> @@ -437,6 +438,55 @@ static int mlx5_devlink_large_group_num_validate(struct devlink *devlink, u32 id
>  	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);
> +	int err = 0;
> +
> +	if (!MLX5_ESWITCH_MANAGER(dev))
> +		return -EOPNOTSUPP;
> +
> +	if (ctx->val.vbool)
> +		err = mlx5_lag_mpesw_enable(dev);
> +	else
> +		mlx5_lag_mpesw_disable(dev);
> +
> +	return err;

How about

	if (ctx->val.vbool)
		return mlx5_lag_mpesw_enable(dev);
	else
		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)
[...]

Thanks,
Olek

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Jakub Kicinski]

On Tue, 14 Feb 2023 09:31:10 +0200 Mark Bloch wrote:
> > Oh, the "legacy software where updating the logic isn't possible"
> > sounds concerning. Do we know what those cases are? Can we perhaps
> > constrain them to run on a specific version of HW (say starting with
> > CX8 the param will no longer be there and only the right behavior will
> > be supported upstream)?  
> 
> While I can encourage customers to update their software to deal with
> this new mode it's up to them. As you already know, it's hard to change
> customers infrastructure so quickly. These things take time. They plan
> on running multiple hardware generations on top of the same software.
> I'll keep pushing on making this mode the default one.

I think we should document some time horizon. We can always push it
back but having a concrete plan in place may motive users. Also it's
useful to point at and say "we warned you". 

Doesn't have to be anything very imminent, time flies.

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Saeed Mahameed]

On 14 Feb 18:07, Alexander Lobakin wrote:
>From: Saeed Mahameed <saeed@kernel.org>
>Date: Fri, 10 Feb 2023 14:18:07 -0800
>

[...]

> +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);
>> +	int err = 0;
>> +
>> +	if (!MLX5_ESWITCH_MANAGER(dev))
>> +		return -EOPNOTSUPP;
>> +
>> +	if (ctx->val.vbool)
>> +		err = mlx5_lag_mpesw_enable(dev);
>> +	else
>> +		mlx5_lag_mpesw_disable(dev);
>> +
>> +	return err;
>
>How about
>
>	if (ctx->val.vbool)
>		return mlx5_lag_mpesw_enable(dev);
>	else
>		mlx5_lag_mpesw_disable(dev);
>
>	return 0;
>
>?

Makes sense, will do the change.

Thanks.

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Saeed Mahameed]

On 14 Feb 12:40, Jakub Kicinski wrote:
>On Tue, 14 Feb 2023 09:31:10 +0200 Mark Bloch wrote:
>> > Oh, the "legacy software where updating the logic isn't possible"
>> > sounds concerning. Do we know what those cases are? Can we perhaps
>> > constrain them to run on a specific version of HW (say starting with
>> > CX8 the param will no longer be there and only the right behavior will
>> > be supported upstream)?
>>
>> While I can encourage customers to update their software to deal with
>> this new mode it's up to them. As you already know, it's hard to change
>> customers infrastructure so quickly. These things take time. They plan
>> on running multiple hardware generations on top of the same software.
>> I'll keep pushing on making this mode the default one.
>
>I think we should document some time horizon. We can always push it
>back but having a concrete plan in place may motive users. Also it's
>useful to point at and say "we warned you".
>
>Doesn't have to be anything very imminent, time flies.

Makes sense, will add to documentation..
Thanks.

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Leon Romanovsky]

On Tue, Feb 14, 2023 at 06:07:54PM +0100, Alexander Lobakin wrote:
> From: Saeed Mahameed <saeed@kernel.org>
> Date: Fri, 10 Feb 2023 14:18:07 -0800
> 
> > From: Roi Dayan <roid@nvidia.com>
> > 
> > Instead of activating multiport eswitch dynamically through
> > adding a TC rule and meeting certain conditions, allow the user
> > to activate it through devlink.
> > This will remove the forced requirement of using TC.
> > e.g. Bridge offload.
> > 
> > Example:
> >     $ devlink dev param set pci/0000:00:0b.0 name esw_multiport value 1 \
> >                   cmode runtime
> > 
> > Signed-off-by: Roi Dayan <roid@nvidia.com>
> > Reviewed-by: Maor Dickman <maord@nvidia.com>
> > Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
> > ---
> >  Documentation/networking/devlink/mlx5.rst     |  4 ++
> >  .../net/ethernet/mellanox/mlx5/core/devlink.c | 56 +++++++++++++++++++
> >  .../net/ethernet/mellanox/mlx5/core/devlink.h |  1 +
> >  .../mellanox/mlx5/core/en/tc/act/mirred.c     |  9 ---
> >  .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 22 +-------
> >  .../net/ethernet/mellanox/mlx5/core/en_tc.h   |  6 --
> >  .../net/ethernet/mellanox/mlx5/core/lag/lag.c |  4 +-
> >  .../net/ethernet/mellanox/mlx5/core/lag/lag.h |  1 +
> >  .../ethernet/mellanox/mlx5/core/lag/mpesw.c   | 46 +++++++--------
> >  .../ethernet/mellanox/mlx5/core/lag/mpesw.h   | 12 +---
> >  10 files changed, 87 insertions(+), 74 deletions(-)
> > 
> > diff --git a/Documentation/networking/devlink/mlx5.rst b/Documentation/networking/devlink/mlx5.rst
> > index 29ad304e6fba..1d2ad2727da1 100644
> > --- a/Documentation/networking/devlink/mlx5.rst
> > +++ b/Documentation/networking/devlink/mlx5.rst
> > @@ -54,6 +54,10 @@ parameters.
> >       - Control the number of large groups (size > 1) in the FDB table.
> >  
> >         * The default value is 15, and the range is between 1 and 1024.
> > +   * - ``esw_multiport``
> > +     - Boolean
> > +     - runtime
> > +     - Set the E-Switch lag mode to multiport.
> >  
> >  The ``mlx5`` driver supports reloading via ``DEVLINK_CMD_RELOAD``
> >  
> > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
> > index b742e04deec1..49392870f695 100644
> > --- a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
> > +++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
> > @@ -7,6 +7,7 @@
> >  #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"
> > @@ -437,6 +438,55 @@ static int mlx5_devlink_large_group_num_validate(struct devlink *devlink, u32 id
> >  	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);
> > +	int err = 0;
> > +
> > +	if (!MLX5_ESWITCH_MANAGER(dev))
> > +		return -EOPNOTSUPP;
> > +
> > +	if (ctx->val.vbool)
> > +		err = mlx5_lag_mpesw_enable(dev);
> > +	else
> > +		mlx5_lag_mpesw_disable(dev);
> > +
> > +	return err;
> 
> How about
> 
> 	if (ctx->val.vbool)
> 		return mlx5_lag_mpesw_enable(dev);
> 	else
> 		mlx5_lag_mpesw_disable(dev);
> 
> 	return 0;

If such construction is used, there won't need in "else".

 	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)
> [...]
> 
> Thanks,
> Olek

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Alexander Lobakin]

From: Leon Romanovsky <leon@kernel.org>
Date: Wed, 15 Feb 2023 13:46:29 +0200

> On Tue, Feb 14, 2023 at 06:07:54PM +0100, Alexander Lobakin wrote:
>> From: Saeed Mahameed <saeed@kernel.org>
>> Date: Fri, 10 Feb 2023 14:18:07 -0800

[...]

>>> @@ -437,6 +438,55 @@ static int mlx5_devlink_large_group_num_validate(struct devlink *devlink, u32 id
>>>  	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);
>>> +	int err = 0;
>>> +
>>> +	if (!MLX5_ESWITCH_MANAGER(dev))
>>> +		return -EOPNOTSUPP;
>>> +
>>> +	if (ctx->val.vbool)
>>> +		err = mlx5_lag_mpesw_enable(dev);
>>> +	else
>>> +		mlx5_lag_mpesw_disable(dev);
>>> +
>>> +	return err;
>>
>> How about
>>
>> 	if (ctx->val.vbool)
>> 		return mlx5_lag_mpesw_enable(dev);
>> 	else
>> 		mlx5_lag_mpesw_disable(dev);
>>
>> 	return 0;
> 
> If such construction is used, there won't need in "else".
> 
>  	if (ctx->val.vbool)
>  		return mlx5_lag_mpesw_enable(dev);
> 
>  	mlx5_lag_mpesw_disable(dev);
>  	return 0;

Correct, I just thought that if-else would look more intuitive here
since it's a simple "if enabled enable else disable".

[...]

Thanks,
Olek

Re: [net-next 01/15] net/mlx5: Lag, Let user configure multiport eswitch

[Saeed Mahameed]

On 15 Feb 18:04, Alexander Lobakin wrote:
>From: Leon Romanovsky <leon@kernel.org>
>Date: Wed, 15 Feb 2023 13:46:29 +0200
>
>> On Tue, Feb 14, 2023 at 06:07:54PM +0100, Alexander Lobakin wrote:
>>> From: Saeed Mahameed <saeed@kernel.org>
>>> Date: Fri, 10 Feb 2023 14:18:07 -0800
>
>[...]
>
>>> How about
>>>
>>> 	if (ctx->val.vbool)
>>> 		return mlx5_lag_mpesw_enable(dev);
>>> 	else
>>> 		mlx5_lag_mpesw_disable(dev);
>>>
>>> 	return 0;
>>
>> If such construction is used, there won't need in "else".
>>
>>  	if (ctx->val.vbool)
>>  		return mlx5_lag_mpesw_enable(dev);
>>
>>  	mlx5_lag_mpesw_disable(dev);
>>  	return 0;
>

Thanks, this is exactly what I did when posting V2.

>Correct, I just thought that if-else would look more intuitive here
>since it's a simple "if enabled enable else disable".
>
>[...]
>
>Thanks,
>Olek