[PATCH v4 net-next 00/11] FDB entries on DSA LAG interfaces

[PATCH v4 net-next 00/11] FDB entries on DSA LAG interfaces

[Vladimir Oltean]

v3->v4:
- avoid NULL pointer dereference in dsa_port_lag_leave() when the LAG is
  not offloaded (thanks to Alvin Šipraga)
- remove the "void *ctx" left over in struct dsa_switchdev_event_work
- make sure the dp->lag assignment is last in dsa_port_lag_create()
v2->v3:
- Move the complexity of iterating over DSA slave interfaces that are
  members of the LAG bridge port from dsa_slave_fdb_event() to
  switchdev_handle_fdb_event_to_device().

This work permits having static and local FDB entries on LAG interfaces
that are offloaded by DSA ports. New API needs to be introduced in
drivers. To maintain consistency with the bridging offload code, I've
taken the liberty to reorganize the data structures added by Tobias in
the DSA core a little bit.

Tested on NXP LS1028A (felix switch). Would appreciate feedback/testing
on other platforms too. Testing procedure was the one described here:
https://patchwork.kernel.org/project/netdevbpf/cover/20210205130240.4072854-1-vladimir.oltean@nxp.com/

with this script:

ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge && ip link set br0 up
ip link set br0 arp off
ip link set bond0 master br0 && ip link set bond0 up
ip link set swp0 master br0 && ip link set swp0 up
ip link set dev bond0 type bridge_slave flood off learning off
bridge fdb add dev bond0 <mac address of other eno0> master static

I'm noticing a problem in 'bridge fdb dump' with the 'self' entries, and
I didn't solve this. On Ocelot, an entry learned on a LAG is reported as
being on the first member port of it (so instead of saying 'self bond0',
it says 'self swp1'). This is better than not seeing the entry at all,
but when DSA queries for the FDBs on a port via ds->ops->port_fdb_dump,
it never queries for FDBs on a LAG. Not clear what we should do there,
we aren't in control of the ->ndo_fdb_dump of the bonding/team drivers.
Alternatively, we could just consider the 'self' entries reported via
ndo_fdb_dump as "better than nothing", and concentrate on the 'master'
entries that are in sync with the bridge when packets are flooded to
software.

Vladimir Oltean (11):
  net: dsa: rename references to "lag" as "lag_dev"
  net: dsa: mv88e6xxx: rename references to "lag" as "lag_dev"
  net: dsa: qca8k: rename references to "lag" as "lag_dev"
  net: dsa: make LAG IDs one-based
  net: dsa: mv88e6xxx: use dsa_switch_for_each_port in
    mv88e6xxx_lag_sync_masks
  net: dsa: create a dsa_lag structure
  net: switchdev: remove lag_mod_cb from
    switchdev_handle_fdb_event_to_device
  net: dsa: remove "ds" and "port" from struct dsa_switchdev_event_work
  net: dsa: call SWITCHDEV_FDB_OFFLOADED for the orig_dev
  net: dsa: support FDB events on offloaded LAG interfaces
  net: dsa: felix: support FDB entries on offloaded LAG interfaces

 drivers/net/dsa/mv88e6xxx/chip.c              |  46 ++++---
 drivers/net/dsa/ocelot/felix.c                |  26 +++-
 drivers/net/dsa/qca8k.c                       |  32 ++---
 .../microchip/lan966x/lan966x_switchdev.c     |  12 +-
 drivers/net/ethernet/mscc/ocelot.c            | 128 +++++++++++++++++-
 include/net/dsa.h                             |  66 ++++++---
 include/net/switchdev.h                       |  10 +-
 include/soc/mscc/ocelot.h                     |  12 ++
 net/dsa/dsa2.c                                |  45 +++---
 net/dsa/dsa_priv.h                            |  24 +++-
 net/dsa/port.c                                |  97 ++++++++++---
 net/dsa/slave.c                               |  64 +++++----
 net/dsa/switch.c                              | 109 +++++++++++++++
 net/dsa/tag_dsa.c                             |   4 +-
 net/switchdev/switchdev.c                     |  80 ++++-------
 15 files changed, 560 insertions(+), 195 deletions(-)

-- 
2.25.1

[PATCH v4 net-next 01/11] net: dsa: rename references to "lag" as "lag_dev"

[Vladimir Oltean]

In preparation of converting struct net_device *dp->lag_dev into a
struct dsa_lag *dp->lag, we need to rename, for consistency purposes,
all occurrences of the "lag" variable in the DSA core to "lag_dev".

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v4: none

 include/net/dsa.h  | 12 ++++++------
 net/dsa/dsa2.c     | 16 ++++++++--------
 net/dsa/dsa_priv.h |  6 +++---
 net/dsa/port.c     | 20 ++++++++++----------
 net/dsa/switch.c   |  8 ++++----
 5 files changed, 31 insertions(+), 31 deletions(-)

diff --git a/include/net/dsa.h b/include/net/dsa.h
index f13de2d8aef3..ef7f446cbdf4 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -182,12 +182,12 @@ static inline struct net_device *dsa_lag_dev(struct dsa_switch_tree *dst,
 }
 
 static inline int dsa_lag_id(struct dsa_switch_tree *dst,
-			     struct net_device *lag)
+			     struct net_device *lag_dev)
 {
 	unsigned int id;
 
 	dsa_lags_foreach_id(id, dst) {
-		if (dsa_lag_dev(dst, id) == lag)
+		if (dsa_lag_dev(dst, id) == lag_dev)
 			return id;
 	}
 
@@ -966,10 +966,10 @@ struct dsa_switch_ops {
 	int	(*crosschip_lag_change)(struct dsa_switch *ds, int sw_index,
 					int port);
 	int	(*crosschip_lag_join)(struct dsa_switch *ds, int sw_index,
-				      int port, struct net_device *lag,
+				      int port, struct net_device *lag_dev,
 				      struct netdev_lag_upper_info *info);
 	int	(*crosschip_lag_leave)(struct dsa_switch *ds, int sw_index,
-				       int port, struct net_device *lag);
+				       int port, struct net_device *lag_dev);
 
 	/*
 	 * PTP functionality
@@ -1041,10 +1041,10 @@ struct dsa_switch_ops {
 	 */
 	int	(*port_lag_change)(struct dsa_switch *ds, int port);
 	int	(*port_lag_join)(struct dsa_switch *ds, int port,
-				 struct net_device *lag,
+				 struct net_device *lag_dev,
 				 struct netdev_lag_upper_info *info);
 	int	(*port_lag_leave)(struct dsa_switch *ds, int port,
-				  struct net_device *lag);
+				  struct net_device *lag_dev);
 
 	/*
 	 * HSR integration
diff --git a/net/dsa/dsa2.c b/net/dsa/dsa2.c
index 408b79a28cd4..01a8efcaabac 100644
--- a/net/dsa/dsa2.c
+++ b/net/dsa/dsa2.c
@@ -74,7 +74,7 @@ int dsa_broadcast(unsigned long e, void *v)
 /**
  * dsa_lag_map() - Map LAG netdev to a linear LAG ID
  * @dst: Tree in which to record the mapping.
- * @lag: Netdev that is to be mapped to an ID.
+ * @lag_dev: Netdev that is to be mapped to an ID.
  *
  * dsa_lag_id/dsa_lag_dev can then be used to translate between the
  * two spaces. The size of the mapping space is determined by the
@@ -82,17 +82,17 @@ int dsa_broadcast(unsigned long e, void *v)
  * it unset if it is not needed, in which case these functions become
  * no-ops.
  */
-void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag)
+void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag_dev)
 {
 	unsigned int id;
 
-	if (dsa_lag_id(dst, lag) >= 0)
+	if (dsa_lag_id(dst, lag_dev) >= 0)
 		/* Already mapped */
 		return;
 
 	for (id = 0; id < dst->lags_len; id++) {
 		if (!dsa_lag_dev(dst, id)) {
-			dst->lags[id] = lag;
+			dst->lags[id] = lag_dev;
 			return;
 		}
 	}
@@ -108,22 +108,22 @@ void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag)
 /**
  * dsa_lag_unmap() - Remove a LAG ID mapping
  * @dst: Tree in which the mapping is recorded.
- * @lag: Netdev that was mapped.
+ * @lag_dev: Netdev that was mapped.
  *
  * As there may be multiple users of the mapping, it is only removed
  * if there are no other references to it.
  */
-void dsa_lag_unmap(struct dsa_switch_tree *dst, struct net_device *lag)
+void dsa_lag_unmap(struct dsa_switch_tree *dst, struct net_device *lag_dev)
 {
 	struct dsa_port *dp;
 	unsigned int id;
 
-	dsa_lag_foreach_port(dp, dst, lag)
+	dsa_lag_foreach_port(dp, dst, lag_dev)
 		/* There are remaining users of this mapping */
 		return;
 
 	dsa_lags_foreach_id(id, dst) {
-		if (dsa_lag_dev(dst, id) == lag) {
+		if (dsa_lag_dev(dst, id) == lag_dev) {
 			dst->lags[id] = NULL;
 			break;
 		}
diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
index a37f0883676a..0293a749b3ac 100644
--- a/net/dsa/dsa_priv.h
+++ b/net/dsa/dsa_priv.h
@@ -76,7 +76,7 @@ struct dsa_notifier_mdb_info {
 
 /* DSA_NOTIFIER_LAG_* */
 struct dsa_notifier_lag_info {
-	struct net_device *lag;
+	struct net_device *lag_dev;
 	int sw_index;
 	int port;
 
@@ -487,8 +487,8 @@ int dsa_switch_register_notifier(struct dsa_switch *ds);
 void dsa_switch_unregister_notifier(struct dsa_switch *ds);
 
 /* dsa2.c */
-void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag);
-void dsa_lag_unmap(struct dsa_switch_tree *dst, struct net_device *lag);
+void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag_dev);
+void dsa_lag_unmap(struct dsa_switch_tree *dst, struct net_device *lag_dev);
 int dsa_tree_notify(struct dsa_switch_tree *dst, unsigned long e, void *v);
 int dsa_broadcast(unsigned long e, void *v);
 int dsa_tree_change_tag_proto(struct dsa_switch_tree *dst,
diff --git a/net/dsa/port.c b/net/dsa/port.c
index 8ce5b259ee95..9ee1236bf3b3 100644
--- a/net/dsa/port.c
+++ b/net/dsa/port.c
@@ -447,27 +447,27 @@ int dsa_port_lag_change(struct dsa_port *dp,
 	return dsa_port_notify(dp, DSA_NOTIFIER_LAG_CHANGE, &info);
 }
 
-int dsa_port_lag_join(struct dsa_port *dp, struct net_device *lag,
+int dsa_port_lag_join(struct dsa_port *dp, struct net_device *lag_dev,
 		      struct netdev_lag_upper_info *uinfo,
 		      struct netlink_ext_ack *extack)
 {
 	struct dsa_notifier_lag_info info = {
 		.sw_index = dp->ds->index,
 		.port = dp->index,
-		.lag = lag,
+		.lag_dev = lag_dev,
 		.info = uinfo,
 	};
 	struct net_device *bridge_dev;
 	int err;
 
-	dsa_lag_map(dp->ds->dst, lag);
-	dp->lag_dev = lag;
+	dsa_lag_map(dp->ds->dst, lag_dev);
+	dp->lag_dev = lag_dev;
 
 	err = dsa_port_notify(dp, DSA_NOTIFIER_LAG_JOIN, &info);
 	if (err)
 		goto err_lag_join;
 
-	bridge_dev = netdev_master_upper_dev_get(lag);
+	bridge_dev = netdev_master_upper_dev_get(lag_dev);
 	if (!bridge_dev || !netif_is_bridge_master(bridge_dev))
 		return 0;
 
@@ -481,11 +481,11 @@ int dsa_port_lag_join(struct dsa_port *dp, struct net_device *lag,
 	dsa_port_notify(dp, DSA_NOTIFIER_LAG_LEAVE, &info);
 err_lag_join:
 	dp->lag_dev = NULL;
-	dsa_lag_unmap(dp->ds->dst, lag);
+	dsa_lag_unmap(dp->ds->dst, lag_dev);
 	return err;
 }
 
-void dsa_port_pre_lag_leave(struct dsa_port *dp, struct net_device *lag)
+void dsa_port_pre_lag_leave(struct dsa_port *dp, struct net_device *lag_dev)
 {
 	struct net_device *br = dsa_port_bridge_dev_get(dp);
 
@@ -493,13 +493,13 @@ void dsa_port_pre_lag_leave(struct dsa_port *dp, struct net_device *lag)
 		dsa_port_pre_bridge_leave(dp, br);
 }
 
-void dsa_port_lag_leave(struct dsa_port *dp, struct net_device *lag)
+void dsa_port_lag_leave(struct dsa_port *dp, struct net_device *lag_dev)
 {
 	struct net_device *br = dsa_port_bridge_dev_get(dp);
 	struct dsa_notifier_lag_info info = {
 		.sw_index = dp->ds->index,
 		.port = dp->index,
-		.lag = lag,
+		.lag_dev = lag_dev,
 	};
 	int err;
 
@@ -521,7 +521,7 @@ void dsa_port_lag_leave(struct dsa_port *dp, struct net_device *lag)
 			"port %d failed to notify DSA_NOTIFIER_LAG_LEAVE: %pe\n",
 			dp->index, ERR_PTR(err));
 
-	dsa_lag_unmap(dp->ds->dst, lag);
+	dsa_lag_unmap(dp->ds->dst, lag_dev);
 }
 
 /* Must be called under rcu_read_lock() */
diff --git a/net/dsa/switch.c b/net/dsa/switch.c
index 0bb3987bd4e6..c71bade9269e 100644
--- a/net/dsa/switch.c
+++ b/net/dsa/switch.c
@@ -468,12 +468,12 @@ static int dsa_switch_lag_join(struct dsa_switch *ds,
 			       struct dsa_notifier_lag_info *info)
 {
 	if (ds->index == info->sw_index && ds->ops->port_lag_join)
-		return ds->ops->port_lag_join(ds, info->port, info->lag,
+		return ds->ops->port_lag_join(ds, info->port, info->lag_dev,
 					      info->info);
 
 	if (ds->index != info->sw_index && ds->ops->crosschip_lag_join)
 		return ds->ops->crosschip_lag_join(ds, info->sw_index,
-						   info->port, info->lag,
+						   info->port, info->lag_dev,
 						   info->info);
 
 	return -EOPNOTSUPP;
@@ -483,11 +483,11 @@ static int dsa_switch_lag_leave(struct dsa_switch *ds,
 				struct dsa_notifier_lag_info *info)
 {
 	if (ds->index == info->sw_index && ds->ops->port_lag_leave)
-		return ds->ops->port_lag_leave(ds, info->port, info->lag);
+		return ds->ops->port_lag_leave(ds, info->port, info->lag_dev);
 
 	if (ds->index != info->sw_index && ds->ops->crosschip_lag_leave)
 		return ds->ops->crosschip_lag_leave(ds, info->sw_index,
-						    info->port, info->lag);
+						    info->port, info->lag_dev);
 
 	return -EOPNOTSUPP;
 }
-- 
2.25.1

Re: [PATCH v4 net-next 01/11] net: dsa: rename references to "lag" as "lag_dev"

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> In preparation of converting struct net_device *dp->lag_dev into a
> struct dsa_lag *dp->lag, we need to rename, for consistency purposes,
> all occurrences of the "lag" variable in the DSA core to "lag_dev".
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 02/11] net: dsa: mv88e6xxx: rename references to "lag" as "lag_dev"

[Vladimir Oltean]

In preparation of converting struct net_device *dp->lag_dev into a
struct dsa_lag *dp->lag, we need to rename, for consistency purposes,
all occurrences of the "lag" variable in mv88e6xxx to "lag_dev".

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v4: none

 drivers/net/dsa/mv88e6xxx/chip.c | 49 ++++++++++++++++----------------
 1 file changed, 25 insertions(+), 24 deletions(-)

diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
index 7d5e72cdc125..24da1aa56546 100644
--- a/drivers/net/dsa/mv88e6xxx/chip.c
+++ b/drivers/net/dsa/mv88e6xxx/chip.c
@@ -6168,7 +6168,7 @@ static int mv88e6xxx_port_bridge_flags(struct dsa_switch *ds, int port,
 }
 
 static bool mv88e6xxx_lag_can_offload(struct dsa_switch *ds,
-				      struct net_device *lag,
+				      struct net_device *lag_dev,
 				      struct netdev_lag_upper_info *info)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
@@ -6178,11 +6178,11 @@ static bool mv88e6xxx_lag_can_offload(struct dsa_switch *ds,
 	if (!mv88e6xxx_has_lag(chip))
 		return false;
 
-	id = dsa_lag_id(ds->dst, lag);
+	id = dsa_lag_id(ds->dst, lag_dev);
 	if (id < 0 || id >= ds->num_lag_ids)
 		return false;
 
-	dsa_lag_foreach_port(dp, ds->dst, lag)
+	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
 		/* Includes the port joining the LAG */
 		members++;
 
@@ -6202,20 +6202,21 @@ static bool mv88e6xxx_lag_can_offload(struct dsa_switch *ds,
 	return true;
 }
 
-static int mv88e6xxx_lag_sync_map(struct dsa_switch *ds, struct net_device *lag)
+static int mv88e6xxx_lag_sync_map(struct dsa_switch *ds,
+				  struct net_device *lag_dev)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	struct dsa_port *dp;
 	u16 map = 0;
 	int id;
 
-	id = dsa_lag_id(ds->dst, lag);
+	id = dsa_lag_id(ds->dst, lag_dev);
 
 	/* Build the map of all ports to distribute flows destined for
 	 * this LAG. This can be either a local user port, or a DSA
 	 * port if the LAG port is on a remote chip.
 	 */
-	dsa_lag_foreach_port(dp, ds->dst, lag)
+	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
 		map |= BIT(dsa_towards_port(ds, dp->ds->index, dp->index));
 
 	return mv88e6xxx_g2_trunk_mapping_write(chip, id, map);
@@ -6259,8 +6260,8 @@ static void mv88e6xxx_lag_set_port_mask(u16 *mask, int port,
 static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
+	struct net_device *lag_dev;
 	unsigned int id, num_tx;
-	struct net_device *lag;
 	struct dsa_port *dp;
 	int i, err, nth;
 	u16 mask[8];
@@ -6284,12 +6285,12 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 	 * are in the Tx set.
 	 */
 	dsa_lags_foreach_id(id, ds->dst) {
-		lag = dsa_lag_dev(ds->dst, id);
-		if (!lag)
+		lag_dev = dsa_lag_dev(ds->dst, id);
+		if (!lag_dev)
 			continue;
 
 		num_tx = 0;
-		dsa_lag_foreach_port(dp, ds->dst, lag) {
+		dsa_lag_foreach_port(dp, ds->dst, lag_dev) {
 			if (dp->lag_tx_enabled)
 				num_tx++;
 		}
@@ -6298,7 +6299,7 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 			continue;
 
 		nth = 0;
-		dsa_lag_foreach_port(dp, ds->dst, lag) {
+		dsa_lag_foreach_port(dp, ds->dst, lag_dev) {
 			if (!dp->lag_tx_enabled)
 				continue;
 
@@ -6320,14 +6321,14 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 }
 
 static int mv88e6xxx_lag_sync_masks_map(struct dsa_switch *ds,
-					struct net_device *lag)
+					struct net_device *lag_dev)
 {
 	int err;
 
 	err = mv88e6xxx_lag_sync_masks(ds);
 
 	if (!err)
-		err = mv88e6xxx_lag_sync_map(ds, lag);
+		err = mv88e6xxx_lag_sync_map(ds, lag_dev);
 
 	return err;
 }
@@ -6344,16 +6345,16 @@ static int mv88e6xxx_port_lag_change(struct dsa_switch *ds, int port)
 }
 
 static int mv88e6xxx_port_lag_join(struct dsa_switch *ds, int port,
-				   struct net_device *lag,
+				   struct net_device *lag_dev,
 				   struct netdev_lag_upper_info *info)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err, id;
 
-	if (!mv88e6xxx_lag_can_offload(ds, lag, info))
+	if (!mv88e6xxx_lag_can_offload(ds, lag_dev, info))
 		return -EOPNOTSUPP;
 
-	id = dsa_lag_id(ds->dst, lag);
+	id = dsa_lag_id(ds->dst, lag_dev);
 
 	mv88e6xxx_reg_lock(chip);
 
@@ -6361,7 +6362,7 @@ static int mv88e6xxx_port_lag_join(struct dsa_switch *ds, int port,
 	if (err)
 		goto err_unlock;
 
-	err = mv88e6xxx_lag_sync_masks_map(ds, lag);
+	err = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
 	if (err)
 		goto err_clear_trunk;
 
@@ -6376,13 +6377,13 @@ static int mv88e6xxx_port_lag_join(struct dsa_switch *ds, int port,
 }
 
 static int mv88e6xxx_port_lag_leave(struct dsa_switch *ds, int port,
-				    struct net_device *lag)
+				    struct net_device *lag_dev)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err_sync, err_trunk;
 
 	mv88e6xxx_reg_lock(chip);
-	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag);
+	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
 	err_trunk = mv88e6xxx_port_set_trunk(chip, port, false, 0);
 	mv88e6xxx_reg_unlock(chip);
 	return err_sync ? : err_trunk;
@@ -6401,18 +6402,18 @@ static int mv88e6xxx_crosschip_lag_change(struct dsa_switch *ds, int sw_index,
 }
 
 static int mv88e6xxx_crosschip_lag_join(struct dsa_switch *ds, int sw_index,
-					int port, struct net_device *lag,
+					int port, struct net_device *lag_dev,
 					struct netdev_lag_upper_info *info)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err;
 
-	if (!mv88e6xxx_lag_can_offload(ds, lag, info))
+	if (!mv88e6xxx_lag_can_offload(ds, lag_dev, info))
 		return -EOPNOTSUPP;
 
 	mv88e6xxx_reg_lock(chip);
 
-	err = mv88e6xxx_lag_sync_masks_map(ds, lag);
+	err = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
 	if (err)
 		goto unlock;
 
@@ -6424,13 +6425,13 @@ static int mv88e6xxx_crosschip_lag_join(struct dsa_switch *ds, int sw_index,
 }
 
 static int mv88e6xxx_crosschip_lag_leave(struct dsa_switch *ds, int sw_index,
-					 int port, struct net_device *lag)
+					 int port, struct net_device *lag_dev)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err_sync, err_pvt;
 
 	mv88e6xxx_reg_lock(chip);
-	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag);
+	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
 	err_pvt = mv88e6xxx_pvt_map(chip, sw_index, port);
 	mv88e6xxx_reg_unlock(chip);
 	return err_sync ? : err_pvt;
-- 
2.25.1

Re: [PATCH v4 net-next 02/11] net: dsa: mv88e6xxx: rename references to "lag" as "lag_dev"

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> In preparation of converting struct net_device *dp->lag_dev into a
> struct dsa_lag *dp->lag, we need to rename, for consistency purposes,
> all occurrences of the "lag" variable in mv88e6xxx to "lag_dev".
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 03/11] net: dsa: qca8k: rename references to "lag" as "lag_dev"

[Vladimir Oltean]

In preparation of converting struct net_device *dp->lag_dev into a
struct dsa_lag *dp->lag, we need to rename, for consistency purposes,
all occurrences of the "lag" variable in qca8k to "lag_dev".

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v4: none

 drivers/net/dsa/qca8k.c | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)

diff --git a/drivers/net/dsa/qca8k.c b/drivers/net/dsa/qca8k.c
index 04fa21e37dfa..5691d193aa71 100644
--- a/drivers/net/dsa/qca8k.c
+++ b/drivers/net/dsa/qca8k.c
@@ -2647,17 +2647,17 @@ qca8k_get_tag_protocol(struct dsa_switch *ds, int port,
 
 static bool
 qca8k_lag_can_offload(struct dsa_switch *ds,
-		      struct net_device *lag,
+		      struct net_device *lag_dev,
 		      struct netdev_lag_upper_info *info)
 {
 	struct dsa_port *dp;
 	int id, members = 0;
 
-	id = dsa_lag_id(ds->dst, lag);
+	id = dsa_lag_id(ds->dst, lag_dev);
 	if (id < 0 || id >= ds->num_lag_ids)
 		return false;
 
-	dsa_lag_foreach_port(dp, ds->dst, lag)
+	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
 		/* Includes the port joining the LAG */
 		members++;
 
@@ -2676,7 +2676,7 @@ qca8k_lag_can_offload(struct dsa_switch *ds,
 
 static int
 qca8k_lag_setup_hash(struct dsa_switch *ds,
-		     struct net_device *lag,
+		     struct net_device *lag_dev,
 		     struct netdev_lag_upper_info *info)
 {
 	struct qca8k_priv *priv = ds->priv;
@@ -2684,7 +2684,7 @@ qca8k_lag_setup_hash(struct dsa_switch *ds,
 	u32 hash = 0;
 	int i, id;
 
-	id = dsa_lag_id(ds->dst, lag);
+	id = dsa_lag_id(ds->dst, lag_dev);
 
 	switch (info->hash_type) {
 	case NETDEV_LAG_HASH_L23:
@@ -2716,7 +2716,7 @@ qca8k_lag_setup_hash(struct dsa_switch *ds,
 	if (unique_lag) {
 		priv->lag_hash_mode = hash;
 	} else if (priv->lag_hash_mode != hash) {
-		netdev_err(lag, "Error: Mismatched Hash Mode across different lag is not supported\n");
+		netdev_err(lag_dev, "Error: Mismatched Hash Mode across different lag is not supported\n");
 		return -EOPNOTSUPP;
 	}
 
@@ -2726,13 +2726,13 @@ qca8k_lag_setup_hash(struct dsa_switch *ds,
 
 static int
 qca8k_lag_refresh_portmap(struct dsa_switch *ds, int port,
-			  struct net_device *lag, bool delete)
+			  struct net_device *lag_dev, bool delete)
 {
 	struct qca8k_priv *priv = ds->priv;
 	int ret, id, i;
 	u32 val;
 
-	id = dsa_lag_id(ds->dst, lag);
+	id = dsa_lag_id(ds->dst, lag_dev);
 
 	/* Read current port member */
 	ret = regmap_read(priv->regmap, QCA8K_REG_GOL_TRUNK_CTRL0, &val);
@@ -2795,26 +2795,26 @@ qca8k_lag_refresh_portmap(struct dsa_switch *ds, int port,
 
 static int
 qca8k_port_lag_join(struct dsa_switch *ds, int port,
-		    struct net_device *lag,
+		    struct net_device *lag_dev,
 		    struct netdev_lag_upper_info *info)
 {
 	int ret;
 
-	if (!qca8k_lag_can_offload(ds, lag, info))
+	if (!qca8k_lag_can_offload(ds, lag_dev, info))
 		return -EOPNOTSUPP;
 
-	ret = qca8k_lag_setup_hash(ds, lag, info);
+	ret = qca8k_lag_setup_hash(ds, lag_dev, info);
 	if (ret)
 		return ret;
 
-	return qca8k_lag_refresh_portmap(ds, port, lag, false);
+	return qca8k_lag_refresh_portmap(ds, port, lag_dev, false);
 }
 
 static int
 qca8k_port_lag_leave(struct dsa_switch *ds, int port,
-		     struct net_device *lag)
+		     struct net_device *lag_dev)
 {
-	return qca8k_lag_refresh_portmap(ds, port, lag, true);
+	return qca8k_lag_refresh_portmap(ds, port, lag_dev, true);
 }
 
 static void
-- 
2.25.1

Re: [PATCH v4 net-next 03/11] net: dsa: qca8k: rename references to "lag" as "lag_dev"

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> In preparation of converting struct net_device *dp->lag_dev into a
> struct dsa_lag *dp->lag, we need to rename, for consistency purposes,
> all occurrences of the "lag" variable in qca8k to "lag_dev".
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 04/11] net: dsa: make LAG IDs one-based

[Vladimir Oltean]

The DSA LAG API will be changed to become more similar with the bridge
data structures, where struct dsa_bridge holds an unsigned int num,
which is generated by DSA and is one-based. We have a similar thing
going with the DSA LAG, except that isn't stored anywhere, it is
calculated dynamically by dsa_lag_id() by iterating through dst->lags.

The idea of encoding an invalid (or not requested) LAG ID as zero for
the purpose of simplifying checks in drivers means that the LAG IDs
passed by DSA to drivers need to be one-based too. So back-and-forth
conversion is needed when indexing the dst->lags array, as well as in
drivers which assume a zero-based index.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v4: none

 drivers/net/dsa/mv88e6xxx/chip.c | 13 ++++++++-----
 drivers/net/dsa/qca8k.c          |  5 +++--
 include/net/dsa.h                |  8 +++++---
 net/dsa/dsa2.c                   |  8 ++++----
 net/dsa/tag_dsa.c                |  2 +-
 5 files changed, 21 insertions(+), 15 deletions(-)

diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
index 24da1aa56546..a4f2e9b65d4e 100644
--- a/drivers/net/dsa/mv88e6xxx/chip.c
+++ b/drivers/net/dsa/mv88e6xxx/chip.c
@@ -1630,10 +1630,11 @@ static int mv88e6xxx_pvt_map(struct mv88e6xxx_chip *chip, int dev, int port)
 			 * FORWARD frames, which use the LAG ID as the
 			 * source port, we must translate dev/port to
 			 * the special "LAG device" in the PVT, using
-			 * the LAG ID as the port number.
+			 * the LAG ID (one-based) as the port number
+			 * (zero-based).
 			 */
 			dev = MV88E6XXX_G2_PVT_ADDR_DEV_TRUNK;
-			port = dsa_lag_id(dst, dp->lag_dev);
+			port = dsa_lag_id(dst, dp->lag_dev) - 1;
 		}
 	}
 
@@ -6179,7 +6180,7 @@ static bool mv88e6xxx_lag_can_offload(struct dsa_switch *ds,
 		return false;
 
 	id = dsa_lag_id(ds->dst, lag_dev);
-	if (id < 0 || id >= ds->num_lag_ids)
+	if (id <= 0 || id > ds->num_lag_ids)
 		return false;
 
 	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
@@ -6210,7 +6211,8 @@ static int mv88e6xxx_lag_sync_map(struct dsa_switch *ds,
 	u16 map = 0;
 	int id;
 
-	id = dsa_lag_id(ds->dst, lag_dev);
+	/* DSA LAG IDs are one-based, hardware is zero-based */
+	id = dsa_lag_id(ds->dst, lag_dev) - 1;
 
 	/* Build the map of all ports to distribute flows destined for
 	 * this LAG. This can be either a local user port, or a DSA
@@ -6354,7 +6356,8 @@ static int mv88e6xxx_port_lag_join(struct dsa_switch *ds, int port,
 	if (!mv88e6xxx_lag_can_offload(ds, lag_dev, info))
 		return -EOPNOTSUPP;
 
-	id = dsa_lag_id(ds->dst, lag_dev);
+	/* DSA LAG IDs are one-based */
+	id = dsa_lag_id(ds->dst, lag_dev) - 1;
 
 	mv88e6xxx_reg_lock(chip);
 
diff --git a/drivers/net/dsa/qca8k.c b/drivers/net/dsa/qca8k.c
index 5691d193aa71..ed55e9357c2b 100644
--- a/drivers/net/dsa/qca8k.c
+++ b/drivers/net/dsa/qca8k.c
@@ -2654,7 +2654,7 @@ qca8k_lag_can_offload(struct dsa_switch *ds,
 	int id, members = 0;
 
 	id = dsa_lag_id(ds->dst, lag_dev);
-	if (id < 0 || id >= ds->num_lag_ids)
+	if (id <= 0 || id > ds->num_lag_ids)
 		return false;
 
 	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
@@ -2732,7 +2732,8 @@ qca8k_lag_refresh_portmap(struct dsa_switch *ds, int port,
 	int ret, id, i;
 	u32 val;
 
-	id = dsa_lag_id(ds->dst, lag_dev);
+	/* DSA LAG IDs are one-based, hardware is zero-based */
+	id = dsa_lag_id(ds->dst, lag_dev) - 1;
 
 	/* Read current port member */
 	ret = regmap_read(priv->regmap, QCA8K_REG_GOL_TRUNK_CTRL0, &val);
diff --git a/include/net/dsa.h b/include/net/dsa.h
index ef7f446cbdf4..3ae93adda25d 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -163,9 +163,10 @@ struct dsa_switch_tree {
 	unsigned int last_switch;
 };
 
+/* LAG IDs are one-based, the dst->lags array is zero-based */
 #define dsa_lags_foreach_id(_id, _dst)				\
-	for ((_id) = 0; (_id) < (_dst)->lags_len; (_id)++)	\
-		if ((_dst)->lags[(_id)])
+	for ((_id) = 1; (_id) <= (_dst)->lags_len; (_id)++)	\
+		if ((_dst)->lags[(_id) - 1])
 
 #define dsa_lag_foreach_port(_dp, _dst, _lag)			\
 	list_for_each_entry((_dp), &(_dst)->ports, list)	\
@@ -178,7 +179,8 @@ struct dsa_switch_tree {
 static inline struct net_device *dsa_lag_dev(struct dsa_switch_tree *dst,
 					     unsigned int id)
 {
-	return dst->lags[id];
+	/* DSA LAG IDs are one-based, dst->lags is zero-based */
+	return dst->lags[id - 1];
 }
 
 static inline int dsa_lag_id(struct dsa_switch_tree *dst,
diff --git a/net/dsa/dsa2.c b/net/dsa/dsa2.c
index 01a8efcaabac..4915abe0d4d2 100644
--- a/net/dsa/dsa2.c
+++ b/net/dsa/dsa2.c
@@ -86,13 +86,13 @@ void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag_dev)
 {
 	unsigned int id;
 
-	if (dsa_lag_id(dst, lag_dev) >= 0)
+	if (dsa_lag_id(dst, lag_dev) > 0)
 		/* Already mapped */
 		return;
 
-	for (id = 0; id < dst->lags_len; id++) {
+	for (id = 1; id <= dst->lags_len; id++) {
 		if (!dsa_lag_dev(dst, id)) {
-			dst->lags[id] = lag_dev;
+			dst->lags[id - 1] = lag_dev;
 			return;
 		}
 	}
@@ -124,7 +124,7 @@ void dsa_lag_unmap(struct dsa_switch_tree *dst, struct net_device *lag_dev)
 
 	dsa_lags_foreach_id(id, dst) {
 		if (dsa_lag_dev(dst, id) == lag_dev) {
-			dst->lags[id] = NULL;
+			dst->lags[id - 1] = NULL;
 			break;
 		}
 	}
diff --git a/net/dsa/tag_dsa.c b/net/dsa/tag_dsa.c
index 8abf39dcac64..26435bc4a098 100644
--- a/net/dsa/tag_dsa.c
+++ b/net/dsa/tag_dsa.c
@@ -251,7 +251,7 @@ static struct sk_buff *dsa_rcv_ll(struct sk_buff *skb, struct net_device *dev,
 		 * so we inject the frame directly on the upper
 		 * team/bond.
 		 */
-		skb->dev = dsa_lag_dev(cpu_dp->dst, source_port);
+		skb->dev = dsa_lag_dev(cpu_dp->dst, source_port + 1);
 	} else {
 		skb->dev = dsa_master_find_slave(dev, source_device,
 						 source_port);
-- 
2.25.1

Re: [PATCH v4 net-next 04/11] net: dsa: make LAG IDs one-based

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> The DSA LAG API will be changed to become more similar with the bridge
> data structures, where struct dsa_bridge holds an unsigned int num,
> which is generated by DSA and is one-based. We have a similar thing
> going with the DSA LAG, except that isn't stored anywhere, it is
> calculated dynamically by dsa_lag_id() by iterating through dst->lags.
> 
> The idea of encoding an invalid (or not requested) LAG ID as zero for
> the purpose of simplifying checks in drivers means that the LAG IDs
> passed by DSA to drivers need to be one-based too. So back-and-forth
> conversion is needed when indexing the dst->lags array, as well as in
> drivers which assume a zero-based index.
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 05/11] net: dsa: mv88e6xxx: use dsa_switch_for_each_port in mv88e6xxx_lag_sync_masks

[Vladimir Oltean]

Make the intent of the code more clear by using the dedicated helper for
iterating over the ports of a switch.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v4: none

 drivers/net/dsa/mv88e6xxx/chip.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
index a4f2e9b65d4e..d96db9825033 100644
--- a/drivers/net/dsa/mv88e6xxx/chip.c
+++ b/drivers/net/dsa/mv88e6xxx/chip.c
@@ -6273,8 +6273,8 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 	ivec = BIT(mv88e6xxx_num_ports(chip)) - 1;
 
 	/* Disable all masks for ports that _are_ members of a LAG. */
-	list_for_each_entry(dp, &ds->dst->ports, list) {
-		if (!dp->lag_dev || dp->ds != ds)
+	dsa_switch_for_each_port(dp, ds) {
+		if (!dp->lag_dev)
 			continue;
 
 		ivec &= ~BIT(dp->index);
-- 
2.25.1

Re: [PATCH v4 net-next 05/11] net: dsa: mv88e6xxx: use dsa_switch_for_each_port in mv88e6xxx_lag_sync_masks

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> Make the intent of the code more clear by using the dedicated helper for
> iterating over the ports of a switch.
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 06/11] net: dsa: create a dsa_lag structure

[Vladimir Oltean]

The main purpose of this change is to create a data structure for a LAG
as seen by DSA. This is similar to what we have for bridging - we pass a
copy of this structure by value to ->port_lag_join and ->port_lag_leave.
For now we keep the lag_dev, id and a reference count in it. Future
patches will add a list of FDB entries for the LAG (these also need to
be refcounted to work properly).

The LAG structure is created using dsa_port_lag_create() and destroyed
using dsa_port_lag_destroy(), just like we have for bridging.

Because now, the dsa_lag itself is refcounted, we can simplify
dsa_lag_map() and dsa_lag_unmap(). These functions need to keep a LAG in
the dst->lags array only as long as at least one port uses it. The
refcounting logic inside those functions can be removed now - they are
called only when we should perform the operation.

dsa_lag_dev() is renamed to dsa_lag_by_id() and now returns the dsa_lag
structure instead of the lag_dev net_device.

dsa_lag_foreach_port() now takes the dsa_lag structure as argument.

dst->lags holds an array of dsa_lag structures.

dsa_lag_map() now also saves the dsa_lag->id value, so that linear
walking of dst->lags in drivers using dsa_lag_id() is no longer
necessary. They can just look at lag.id.

dsa_port_lag_id_get() is a helper, similar to dsa_port_bridge_num_get(),
which can be used by drivers to get the LAG ID assigned by DSA to a
given port.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v3->v4: avoid NULL pointer dereference in dsa_port_lag_leave() when the
        LAG is not offloaded (thanks to Alvin Šipraga)
v1->v3: none

 drivers/net/dsa/mv88e6xxx/chip.c | 60 +++++++++++++++----------------
 drivers/net/dsa/ocelot/felix.c   |  8 ++---
 drivers/net/dsa/qca8k.c          | 37 +++++++++----------
 include/net/dsa.h                | 50 +++++++++++++++++++-------
 net/dsa/dsa2.c                   | 41 +++++++++++----------
 net/dsa/dsa_priv.h               |  8 +++--
 net/dsa/port.c                   | 62 +++++++++++++++++++++++++-------
 net/dsa/slave.c                  |  4 +--
 net/dsa/switch.c                 |  8 ++---
 net/dsa/tag_dsa.c                |  4 ++-
 10 files changed, 173 insertions(+), 109 deletions(-)

diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
index d96db9825033..da7b38b1a3b8 100644
--- a/drivers/net/dsa/mv88e6xxx/chip.c
+++ b/drivers/net/dsa/mv88e6xxx/chip.c
@@ -1625,7 +1625,7 @@ static int mv88e6xxx_pvt_map(struct mv88e6xxx_chip *chip, int dev, int port)
 
 		ds = dsa_switch_find(dst->index, dev);
 		dp = ds ? dsa_to_port(ds, port) : NULL;
-		if (dp && dp->lag_dev) {
+		if (dp && dp->lag) {
 			/* As the PVT is used to limit flooding of
 			 * FORWARD frames, which use the LAG ID as the
 			 * source port, we must translate dev/port to
@@ -1634,7 +1634,7 @@ static int mv88e6xxx_pvt_map(struct mv88e6xxx_chip *chip, int dev, int port)
 			 * (zero-based).
 			 */
 			dev = MV88E6XXX_G2_PVT_ADDR_DEV_TRUNK;
-			port = dsa_lag_id(dst, dp->lag_dev) - 1;
+			port = dsa_port_lag_id_get(dp) - 1;
 		}
 	}
 
@@ -1672,7 +1672,7 @@ static void mv88e6xxx_port_fast_age(struct dsa_switch *ds, int port)
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err;
 
-	if (dsa_to_port(ds, port)->lag_dev)
+	if (dsa_to_port(ds, port)->lag)
 		/* Hardware is incapable of fast-aging a LAG through a
 		 * regular ATU move operation. Until we have something
 		 * more fancy in place this is a no-op.
@@ -6169,21 +6169,20 @@ static int mv88e6xxx_port_bridge_flags(struct dsa_switch *ds, int port,
 }
 
 static bool mv88e6xxx_lag_can_offload(struct dsa_switch *ds,
-				      struct net_device *lag_dev,
+				      struct dsa_lag lag,
 				      struct netdev_lag_upper_info *info)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	struct dsa_port *dp;
-	int id, members = 0;
+	int members = 0;
 
 	if (!mv88e6xxx_has_lag(chip))
 		return false;
 
-	id = dsa_lag_id(ds->dst, lag_dev);
-	if (id <= 0 || id > ds->num_lag_ids)
+	if (!lag.id)
 		return false;
 
-	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
+	dsa_lag_foreach_port(dp, ds->dst, &lag)
 		/* Includes the port joining the LAG */
 		members++;
 
@@ -6203,8 +6202,7 @@ static bool mv88e6xxx_lag_can_offload(struct dsa_switch *ds,
 	return true;
 }
 
-static int mv88e6xxx_lag_sync_map(struct dsa_switch *ds,
-				  struct net_device *lag_dev)
+static int mv88e6xxx_lag_sync_map(struct dsa_switch *ds, struct dsa_lag lag)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	struct dsa_port *dp;
@@ -6212,13 +6210,13 @@ static int mv88e6xxx_lag_sync_map(struct dsa_switch *ds,
 	int id;
 
 	/* DSA LAG IDs are one-based, hardware is zero-based */
-	id = dsa_lag_id(ds->dst, lag_dev) - 1;
+	id = lag.id - 1;
 
 	/* Build the map of all ports to distribute flows destined for
 	 * this LAG. This can be either a local user port, or a DSA
 	 * port if the LAG port is on a remote chip.
 	 */
-	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
+	dsa_lag_foreach_port(dp, ds->dst, &lag)
 		map |= BIT(dsa_towards_port(ds, dp->ds->index, dp->index));
 
 	return mv88e6xxx_g2_trunk_mapping_write(chip, id, map);
@@ -6262,9 +6260,9 @@ static void mv88e6xxx_lag_set_port_mask(u16 *mask, int port,
 static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
-	struct net_device *lag_dev;
 	unsigned int id, num_tx;
 	struct dsa_port *dp;
+	struct dsa_lag *lag;
 	int i, err, nth;
 	u16 mask[8];
 	u16 ivec;
@@ -6274,7 +6272,7 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 
 	/* Disable all masks for ports that _are_ members of a LAG. */
 	dsa_switch_for_each_port(dp, ds) {
-		if (!dp->lag_dev)
+		if (!dp->lag)
 			continue;
 
 		ivec &= ~BIT(dp->index);
@@ -6287,12 +6285,12 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 	 * are in the Tx set.
 	 */
 	dsa_lags_foreach_id(id, ds->dst) {
-		lag_dev = dsa_lag_dev(ds->dst, id);
-		if (!lag_dev)
+		lag = dsa_lag_by_id(ds->dst, id);
+		if (!lag)
 			continue;
 
 		num_tx = 0;
-		dsa_lag_foreach_port(dp, ds->dst, lag_dev) {
+		dsa_lag_foreach_port(dp, ds->dst, lag) {
 			if (dp->lag_tx_enabled)
 				num_tx++;
 		}
@@ -6301,7 +6299,7 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 			continue;
 
 		nth = 0;
-		dsa_lag_foreach_port(dp, ds->dst, lag_dev) {
+		dsa_lag_foreach_port(dp, ds->dst, lag) {
 			if (!dp->lag_tx_enabled)
 				continue;
 
@@ -6323,14 +6321,14 @@ static int mv88e6xxx_lag_sync_masks(struct dsa_switch *ds)
 }
 
 static int mv88e6xxx_lag_sync_masks_map(struct dsa_switch *ds,
-					struct net_device *lag_dev)
+					struct dsa_lag lag)
 {
 	int err;
 
 	err = mv88e6xxx_lag_sync_masks(ds);
 
 	if (!err)
-		err = mv88e6xxx_lag_sync_map(ds, lag_dev);
+		err = mv88e6xxx_lag_sync_map(ds, lag);
 
 	return err;
 }
@@ -6347,17 +6345,17 @@ static int mv88e6xxx_port_lag_change(struct dsa_switch *ds, int port)
 }
 
 static int mv88e6xxx_port_lag_join(struct dsa_switch *ds, int port,
-				   struct net_device *lag_dev,
+				   struct dsa_lag lag,
 				   struct netdev_lag_upper_info *info)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err, id;
 
-	if (!mv88e6xxx_lag_can_offload(ds, lag_dev, info))
+	if (!mv88e6xxx_lag_can_offload(ds, lag, info))
 		return -EOPNOTSUPP;
 
 	/* DSA LAG IDs are one-based */
-	id = dsa_lag_id(ds->dst, lag_dev) - 1;
+	id = lag.id - 1;
 
 	mv88e6xxx_reg_lock(chip);
 
@@ -6365,7 +6363,7 @@ static int mv88e6xxx_port_lag_join(struct dsa_switch *ds, int port,
 	if (err)
 		goto err_unlock;
 
-	err = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
+	err = mv88e6xxx_lag_sync_masks_map(ds, lag);
 	if (err)
 		goto err_clear_trunk;
 
@@ -6380,13 +6378,13 @@ static int mv88e6xxx_port_lag_join(struct dsa_switch *ds, int port,
 }
 
 static int mv88e6xxx_port_lag_leave(struct dsa_switch *ds, int port,
-				    struct net_device *lag_dev)
+				    struct dsa_lag lag)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err_sync, err_trunk;
 
 	mv88e6xxx_reg_lock(chip);
-	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
+	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag);
 	err_trunk = mv88e6xxx_port_set_trunk(chip, port, false, 0);
 	mv88e6xxx_reg_unlock(chip);
 	return err_sync ? : err_trunk;
@@ -6405,18 +6403,18 @@ static int mv88e6xxx_crosschip_lag_change(struct dsa_switch *ds, int sw_index,
 }
 
 static int mv88e6xxx_crosschip_lag_join(struct dsa_switch *ds, int sw_index,
-					int port, struct net_device *lag_dev,
+					int port, struct dsa_lag lag,
 					struct netdev_lag_upper_info *info)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err;
 
-	if (!mv88e6xxx_lag_can_offload(ds, lag_dev, info))
+	if (!mv88e6xxx_lag_can_offload(ds, lag, info))
 		return -EOPNOTSUPP;
 
 	mv88e6xxx_reg_lock(chip);
 
-	err = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
+	err = mv88e6xxx_lag_sync_masks_map(ds, lag);
 	if (err)
 		goto unlock;
 
@@ -6428,13 +6426,13 @@ static int mv88e6xxx_crosschip_lag_join(struct dsa_switch *ds, int sw_index,
 }
 
 static int mv88e6xxx_crosschip_lag_leave(struct dsa_switch *ds, int sw_index,
-					 int port, struct net_device *lag_dev)
+					 int port, struct dsa_lag lag)
 {
 	struct mv88e6xxx_chip *chip = ds->priv;
 	int err_sync, err_pvt;
 
 	mv88e6xxx_reg_lock(chip);
-	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag_dev);
+	err_sync = mv88e6xxx_lag_sync_masks_map(ds, lag);
 	err_pvt = mv88e6xxx_pvt_map(chip, sw_index, port);
 	mv88e6xxx_reg_unlock(chip);
 	return err_sync ? : err_pvt;
diff --git a/drivers/net/dsa/ocelot/felix.c b/drivers/net/dsa/ocelot/felix.c
index 9ffd5491bf2d..6d483887af04 100644
--- a/drivers/net/dsa/ocelot/felix.c
+++ b/drivers/net/dsa/ocelot/felix.c
@@ -677,20 +677,20 @@ static void felix_bridge_leave(struct dsa_switch *ds, int port,
 }
 
 static int felix_lag_join(struct dsa_switch *ds, int port,
-			  struct net_device *bond,
+			  struct dsa_lag lag,
 			  struct netdev_lag_upper_info *info)
 {
 	struct ocelot *ocelot = ds->priv;
 
-	return ocelot_port_lag_join(ocelot, port, bond, info);
+	return ocelot_port_lag_join(ocelot, port, lag.dev, info);
 }
 
 static int felix_lag_leave(struct dsa_switch *ds, int port,
-			   struct net_device *bond)
+			   struct dsa_lag lag)
 {
 	struct ocelot *ocelot = ds->priv;
 
-	ocelot_port_lag_leave(ocelot, port, bond);
+	ocelot_port_lag_leave(ocelot, port, lag.dev);
 
 	return 0;
 }
diff --git a/drivers/net/dsa/qca8k.c b/drivers/net/dsa/qca8k.c
index ed55e9357c2b..6844106975a9 100644
--- a/drivers/net/dsa/qca8k.c
+++ b/drivers/net/dsa/qca8k.c
@@ -2646,18 +2646,16 @@ qca8k_get_tag_protocol(struct dsa_switch *ds, int port,
 }
 
 static bool
-qca8k_lag_can_offload(struct dsa_switch *ds,
-		      struct net_device *lag_dev,
+qca8k_lag_can_offload(struct dsa_switch *ds, struct dsa_lag lag,
 		      struct netdev_lag_upper_info *info)
 {
 	struct dsa_port *dp;
-	int id, members = 0;
+	int members = 0;
 
-	id = dsa_lag_id(ds->dst, lag_dev);
-	if (id <= 0 || id > ds->num_lag_ids)
+	if (!lag.id)
 		return false;
 
-	dsa_lag_foreach_port(dp, ds->dst, lag_dev)
+	dsa_lag_foreach_port(dp, ds->dst, &lag)
 		/* Includes the port joining the LAG */
 		members++;
 
@@ -2675,16 +2673,14 @@ qca8k_lag_can_offload(struct dsa_switch *ds,
 }
 
 static int
-qca8k_lag_setup_hash(struct dsa_switch *ds,
-		     struct net_device *lag_dev,
+qca8k_lag_setup_hash(struct dsa_switch *ds, struct dsa_lag lag,
 		     struct netdev_lag_upper_info *info)
 {
+	struct net_device *lag_dev = lag.dev;
 	struct qca8k_priv *priv = ds->priv;
 	bool unique_lag = true;
+	unsigned int i;
 	u32 hash = 0;
-	int i, id;
-
-	id = dsa_lag_id(ds->dst, lag_dev);
 
 	switch (info->hash_type) {
 	case NETDEV_LAG_HASH_L23:
@@ -2701,7 +2697,7 @@ qca8k_lag_setup_hash(struct dsa_switch *ds,
 
 	/* Check if we are the unique configured LAG */
 	dsa_lags_foreach_id(i, ds->dst)
-		if (i != id && dsa_lag_dev(ds->dst, i)) {
+		if (i != lag.id && dsa_lag_by_id(ds->dst, i)) {
 			unique_lag = false;
 			break;
 		}
@@ -2726,14 +2722,14 @@ qca8k_lag_setup_hash(struct dsa_switch *ds,
 
 static int
 qca8k_lag_refresh_portmap(struct dsa_switch *ds, int port,
-			  struct net_device *lag_dev, bool delete)
+			  struct dsa_lag lag, bool delete)
 {
 	struct qca8k_priv *priv = ds->priv;
 	int ret, id, i;
 	u32 val;
 
 	/* DSA LAG IDs are one-based, hardware is zero-based */
-	id = dsa_lag_id(ds->dst, lag_dev) - 1;
+	id = lag.id - 1;
 
 	/* Read current port member */
 	ret = regmap_read(priv->regmap, QCA8K_REG_GOL_TRUNK_CTRL0, &val);
@@ -2795,27 +2791,26 @@ qca8k_lag_refresh_portmap(struct dsa_switch *ds, int port,
 }
 
 static int
-qca8k_port_lag_join(struct dsa_switch *ds, int port,
-		    struct net_device *lag_dev,
+qca8k_port_lag_join(struct dsa_switch *ds, int port, struct dsa_lag lag,
 		    struct netdev_lag_upper_info *info)
 {
 	int ret;
 
-	if (!qca8k_lag_can_offload(ds, lag_dev, info))
+	if (!qca8k_lag_can_offload(ds, lag, info))
 		return -EOPNOTSUPP;
 
-	ret = qca8k_lag_setup_hash(ds, lag_dev, info);
+	ret = qca8k_lag_setup_hash(ds, lag, info);
 	if (ret)
 		return ret;
 
-	return qca8k_lag_refresh_portmap(ds, port, lag_dev, false);
+	return qca8k_lag_refresh_portmap(ds, port, lag, false);
 }
 
 static int
 qca8k_port_lag_leave(struct dsa_switch *ds, int port,
-		     struct net_device *lag_dev)
+		     struct dsa_lag lag)
 {
-	return qca8k_lag_refresh_portmap(ds, port, lag_dev, true);
+	return qca8k_lag_refresh_portmap(ds, port, lag, true);
 }
 
 static void
diff --git a/include/net/dsa.h b/include/net/dsa.h
index 3ae93adda25d..81ed34998416 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -116,6 +116,12 @@ struct dsa_netdevice_ops {
 #define MODULE_ALIAS_DSA_TAG_DRIVER(__proto)				\
 	MODULE_ALIAS(DSA_TAG_DRIVER_ALIAS __stringify(__proto##_VALUE))
 
+struct dsa_lag {
+	struct net_device *dev;
+	unsigned int id;
+	refcount_t refcount;
+};
+
 struct dsa_switch_tree {
 	struct list_head	list;
 
@@ -134,7 +140,7 @@ struct dsa_switch_tree {
 	/* Maps offloaded LAG netdevs to a zero-based linear ID for
 	 * drivers that need it.
 	 */
-	struct net_device **lags;
+	struct dsa_lag **lags;
 
 	/* Tagging protocol operations */
 	const struct dsa_device_ops *tag_ops;
@@ -170,14 +176,14 @@ struct dsa_switch_tree {
 
 #define dsa_lag_foreach_port(_dp, _dst, _lag)			\
 	list_for_each_entry((_dp), &(_dst)->ports, list)	\
-		if ((_dp)->lag_dev == (_lag))
+		if (dsa_port_offloads_lag((_dp), (_lag)))
 
 #define dsa_hsr_foreach_port(_dp, _ds, _hsr)			\
 	list_for_each_entry((_dp), &(_ds)->dst->ports, list)	\
 		if ((_dp)->ds == (_ds) && (_dp)->hsr_dev == (_hsr))
 
-static inline struct net_device *dsa_lag_dev(struct dsa_switch_tree *dst,
-					     unsigned int id)
+static inline struct dsa_lag *dsa_lag_by_id(struct dsa_switch_tree *dst,
+					    unsigned int id)
 {
 	/* DSA LAG IDs are one-based, dst->lags is zero-based */
 	return dst->lags[id - 1];
@@ -189,8 +195,10 @@ static inline int dsa_lag_id(struct dsa_switch_tree *dst,
 	unsigned int id;
 
 	dsa_lags_foreach_id(id, dst) {
-		if (dsa_lag_dev(dst, id) == lag_dev)
-			return id;
+		struct dsa_lag *lag = dsa_lag_by_id(dst, id);
+
+		if (lag->dev == lag_dev)
+			return lag->id;
 	}
 
 	return -ENODEV;
@@ -293,7 +301,7 @@ struct dsa_port {
 	struct devlink_port	devlink_port;
 	struct phylink		*pl;
 	struct phylink_config	pl_config;
-	struct net_device	*lag_dev;
+	struct dsa_lag		*lag;
 	struct net_device	*hsr_dev;
 
 	struct list_head list;
@@ -643,14 +651,30 @@ static inline bool dsa_port_is_vlan_filtering(const struct dsa_port *dp)
 		return dp->vlan_filtering;
 }
 
+static inline unsigned int dsa_port_lag_id_get(struct dsa_port *dp)
+{
+	return dp->lag ? dp->lag->id : 0;
+}
+
+static inline struct net_device *dsa_port_lag_dev_get(struct dsa_port *dp)
+{
+	return dp->lag ? dp->lag->dev : NULL;
+}
+
+static inline bool dsa_port_offloads_lag(struct dsa_port *dp,
+					 const struct dsa_lag *lag)
+{
+	return dsa_port_lag_dev_get(dp) == lag->dev;
+}
+
 static inline
 struct net_device *dsa_port_to_bridge_port(const struct dsa_port *dp)
 {
 	if (!dp->bridge)
 		return NULL;
 
-	if (dp->lag_dev)
-		return dp->lag_dev;
+	if (dp->lag)
+		return dp->lag->dev;
 	else if (dp->hsr_dev)
 		return dp->hsr_dev;
 
@@ -968,10 +992,10 @@ struct dsa_switch_ops {
 	int	(*crosschip_lag_change)(struct dsa_switch *ds, int sw_index,
 					int port);
 	int	(*crosschip_lag_join)(struct dsa_switch *ds, int sw_index,
-				      int port, struct net_device *lag_dev,
+				      int port, struct dsa_lag lag,
 				      struct netdev_lag_upper_info *info);
 	int	(*crosschip_lag_leave)(struct dsa_switch *ds, int sw_index,
-				       int port, struct net_device *lag_dev);
+				       int port, struct dsa_lag lag);
 
 	/*
 	 * PTP functionality
@@ -1043,10 +1067,10 @@ struct dsa_switch_ops {
 	 */
 	int	(*port_lag_change)(struct dsa_switch *ds, int port);
 	int	(*port_lag_join)(struct dsa_switch *ds, int port,
-				 struct net_device *lag_dev,
+				 struct dsa_lag lag,
 				 struct netdev_lag_upper_info *info);
 	int	(*port_lag_leave)(struct dsa_switch *ds, int port,
-				  struct net_device *lag_dev);
+				  struct dsa_lag lag);
 
 	/*
 	 * HSR integration
diff --git a/net/dsa/dsa2.c b/net/dsa/dsa2.c
index 4915abe0d4d2..030d5f26715a 100644
--- a/net/dsa/dsa2.c
+++ b/net/dsa/dsa2.c
@@ -72,27 +72,24 @@ int dsa_broadcast(unsigned long e, void *v)
 }
 
 /**
- * dsa_lag_map() - Map LAG netdev to a linear LAG ID
+ * dsa_lag_map() - Map LAG structure to a linear LAG array
  * @dst: Tree in which to record the mapping.
- * @lag_dev: Netdev that is to be mapped to an ID.
+ * @lag: LAG structure that is to be mapped to the tree's array.
  *
- * dsa_lag_id/dsa_lag_dev can then be used to translate between the
+ * dsa_lag_id/dsa_lag_by_id can then be used to translate between the
  * two spaces. The size of the mapping space is determined by the
  * driver by setting ds->num_lag_ids. It is perfectly legal to leave
  * it unset if it is not needed, in which case these functions become
  * no-ops.
  */
-void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag_dev)
+void dsa_lag_map(struct dsa_switch_tree *dst, struct dsa_lag *lag)
 {
 	unsigned int id;
 
-	if (dsa_lag_id(dst, lag_dev) > 0)
-		/* Already mapped */
-		return;
-
 	for (id = 1; id <= dst->lags_len; id++) {
-		if (!dsa_lag_dev(dst, id)) {
-			dst->lags[id - 1] = lag_dev;
+		if (!dsa_lag_by_id(dst, id)) {
+			dst->lags[id - 1] = lag;
+			lag->id = id;
 			return;
 		}
 	}
@@ -108,28 +105,36 @@ void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag_dev)
 /**
  * dsa_lag_unmap() - Remove a LAG ID mapping
  * @dst: Tree in which the mapping is recorded.
- * @lag_dev: Netdev that was mapped.
+ * @lag: LAG structure that was mapped.
  *
  * As there may be multiple users of the mapping, it is only removed
  * if there are no other references to it.
  */
-void dsa_lag_unmap(struct dsa_switch_tree *dst, struct net_device *lag_dev)
+void dsa_lag_unmap(struct dsa_switch_tree *dst, struct dsa_lag *lag)
 {
-	struct dsa_port *dp;
 	unsigned int id;
 
-	dsa_lag_foreach_port(dp, dst, lag_dev)
-		/* There are remaining users of this mapping */
-		return;
-
 	dsa_lags_foreach_id(id, dst) {
-		if (dsa_lag_dev(dst, id) == lag_dev) {
+		if (dsa_lag_by_id(dst, id) == lag) {
 			dst->lags[id - 1] = NULL;
+			lag->id = 0;
 			break;
 		}
 	}
 }
 
+struct dsa_lag *dsa_tree_lag_find(struct dsa_switch_tree *dst,
+				  const struct net_device *lag_dev)
+{
+	struct dsa_port *dp;
+
+	list_for_each_entry(dp, &dst->ports, list)
+		if (dsa_port_lag_dev_get(dp) == lag_dev)
+			return dp->lag;
+
+	return NULL;
+}
+
 struct dsa_bridge *dsa_tree_bridge_find(struct dsa_switch_tree *dst,
 					const struct net_device *br)
 {
diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
index 0293a749b3ac..8612ff8ea7fe 100644
--- a/net/dsa/dsa_priv.h
+++ b/net/dsa/dsa_priv.h
@@ -76,7 +76,7 @@ struct dsa_notifier_mdb_info {
 
 /* DSA_NOTIFIER_LAG_* */
 struct dsa_notifier_lag_info {
-	struct net_device *lag_dev;
+	struct dsa_lag lag;
 	int sw_index;
 	int port;
 
@@ -487,8 +487,10 @@ int dsa_switch_register_notifier(struct dsa_switch *ds);
 void dsa_switch_unregister_notifier(struct dsa_switch *ds);
 
 /* dsa2.c */
-void dsa_lag_map(struct dsa_switch_tree *dst, struct net_device *lag_dev);
-void dsa_lag_unmap(struct dsa_switch_tree *dst, struct net_device *lag_dev);
+void dsa_lag_map(struct dsa_switch_tree *dst, struct dsa_lag *lag);
+void dsa_lag_unmap(struct dsa_switch_tree *dst, struct dsa_lag *lag);
+struct dsa_lag *dsa_tree_lag_find(struct dsa_switch_tree *dst,
+				  const struct net_device *lag_dev);
 int dsa_tree_notify(struct dsa_switch_tree *dst, unsigned long e, void *v);
 int dsa_broadcast(unsigned long e, void *v);
 int dsa_tree_change_tag_proto(struct dsa_switch_tree *dst,
diff --git a/net/dsa/port.c b/net/dsa/port.c
index 9ee1236bf3b3..bf30ceaf39f4 100644
--- a/net/dsa/port.c
+++ b/net/dsa/port.c
@@ -429,7 +429,7 @@ int dsa_port_lag_change(struct dsa_port *dp,
 	};
 	bool tx_enabled;
 
-	if (!dp->lag_dev)
+	if (!dp->lag)
 		return 0;
 
 	/* On statically configured aggregates (e.g. loadbalance
@@ -447,6 +447,45 @@ int dsa_port_lag_change(struct dsa_port *dp,
 	return dsa_port_notify(dp, DSA_NOTIFIER_LAG_CHANGE, &info);
 }
 
+static int dsa_port_lag_create(struct dsa_port *dp,
+			       struct net_device *lag_dev)
+{
+	struct dsa_switch *ds = dp->ds;
+	struct dsa_lag *lag;
+
+	lag = dsa_tree_lag_find(ds->dst, lag_dev);
+	if (lag) {
+		refcount_inc(&lag->refcount);
+		dp->lag = lag;
+		return 0;
+	}
+
+	lag = kzalloc(sizeof(*lag), GFP_KERNEL);
+	if (!lag)
+		return -ENOMEM;
+
+	refcount_set(&lag->refcount, 1);
+	lag->dev = lag_dev;
+	dsa_lag_map(ds->dst, lag);
+	dp->lag = lag;
+
+	return 0;
+}
+
+static void dsa_port_lag_destroy(struct dsa_port *dp)
+{
+	struct dsa_lag *lag = dp->lag;
+
+	dp->lag = NULL;
+	dp->lag_tx_enabled = false;
+
+	if (!refcount_dec_and_test(&lag->refcount))
+		return;
+
+	dsa_lag_unmap(dp->ds->dst, lag);
+	kfree(lag);
+}
+
 int dsa_port_lag_join(struct dsa_port *dp, struct net_device *lag_dev,
 		      struct netdev_lag_upper_info *uinfo,
 		      struct netlink_ext_ack *extack)
@@ -454,15 +493,16 @@ int dsa_port_lag_join(struct dsa_port *dp, struct net_device *lag_dev,
 	struct dsa_notifier_lag_info info = {
 		.sw_index = dp->ds->index,
 		.port = dp->index,
-		.lag_dev = lag_dev,
 		.info = uinfo,
 	};
 	struct net_device *bridge_dev;
 	int err;
 
-	dsa_lag_map(dp->ds->dst, lag_dev);
-	dp->lag_dev = lag_dev;
+	err = dsa_port_lag_create(dp, lag_dev);
+	if (err)
+		goto err_lag_create;
 
+	info.lag = *dp->lag;
 	err = dsa_port_notify(dp, DSA_NOTIFIER_LAG_JOIN, &info);
 	if (err)
 		goto err_lag_join;
@@ -480,8 +520,8 @@ int dsa_port_lag_join(struct dsa_port *dp, struct net_device *lag_dev,
 err_bridge_join:
 	dsa_port_notify(dp, DSA_NOTIFIER_LAG_LEAVE, &info);
 err_lag_join:
-	dp->lag_dev = NULL;
-	dsa_lag_unmap(dp->ds->dst, lag_dev);
+	dsa_port_lag_destroy(dp);
+err_lag_create:
 	return err;
 }
 
@@ -499,11 +539,10 @@ void dsa_port_lag_leave(struct dsa_port *dp, struct net_device *lag_dev)
 	struct dsa_notifier_lag_info info = {
 		.sw_index = dp->ds->index,
 		.port = dp->index,
-		.lag_dev = lag_dev,
 	};
 	int err;
 
-	if (!dp->lag_dev)
+	if (!dp->lag)
 		return;
 
 	/* Port might have been part of a LAG that in turn was
@@ -512,16 +551,15 @@ void dsa_port_lag_leave(struct dsa_port *dp, struct net_device *lag_dev)
 	if (br)
 		dsa_port_bridge_leave(dp, br);
 
-	dp->lag_tx_enabled = false;
-	dp->lag_dev = NULL;
+	info.lag = *dp->lag;
+
+	dsa_port_lag_destroy(dp);
 
 	err = dsa_port_notify(dp, DSA_NOTIFIER_LAG_LEAVE, &info);
 	if (err)
 		dev_err(dp->ds->dev,
 			"port %d failed to notify DSA_NOTIFIER_LAG_LEAVE: %pe\n",
 			dp->index, ERR_PTR(err));
-
-	dsa_lag_unmap(dp->ds->dst, lag_dev);
 }
 
 /* Must be called under rcu_read_lock() */
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index f61e6b72ffbb..e31c7710fee9 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -2134,7 +2134,7 @@ dsa_slave_lag_changeupper(struct net_device *dev,
 			continue;
 
 		dp = dsa_slave_to_port(lower);
-		if (!dp->lag_dev)
+		if (!dp->lag)
 			/* Software LAG */
 			continue;
 
@@ -2163,7 +2163,7 @@ dsa_slave_lag_prechangeupper(struct net_device *dev,
 			continue;
 
 		dp = dsa_slave_to_port(lower);
-		if (!dp->lag_dev)
+		if (!dp->lag)
 			/* Software LAG */
 			continue;
 
diff --git a/net/dsa/switch.c b/net/dsa/switch.c
index c71bade9269e..0bb3987bd4e6 100644
--- a/net/dsa/switch.c
+++ b/net/dsa/switch.c
@@ -468,12 +468,12 @@ static int dsa_switch_lag_join(struct dsa_switch *ds,
 			       struct dsa_notifier_lag_info *info)
 {
 	if (ds->index == info->sw_index && ds->ops->port_lag_join)
-		return ds->ops->port_lag_join(ds, info->port, info->lag_dev,
+		return ds->ops->port_lag_join(ds, info->port, info->lag,
 					      info->info);
 
 	if (ds->index != info->sw_index && ds->ops->crosschip_lag_join)
 		return ds->ops->crosschip_lag_join(ds, info->sw_index,
-						   info->port, info->lag_dev,
+						   info->port, info->lag,
 						   info->info);
 
 	return -EOPNOTSUPP;
@@ -483,11 +483,11 @@ static int dsa_switch_lag_leave(struct dsa_switch *ds,
 				struct dsa_notifier_lag_info *info)
 {
 	if (ds->index == info->sw_index && ds->ops->port_lag_leave)
-		return ds->ops->port_lag_leave(ds, info->port, info->lag_dev);
+		return ds->ops->port_lag_leave(ds, info->port, info->lag);
 
 	if (ds->index != info->sw_index && ds->ops->crosschip_lag_leave)
 		return ds->ops->crosschip_lag_leave(ds, info->sw_index,
-						    info->port, info->lag_dev);
+						    info->port, info->lag);
 
 	return -EOPNOTSUPP;
 }
diff --git a/net/dsa/tag_dsa.c b/net/dsa/tag_dsa.c
index 26435bc4a098..c8b4bbd46191 100644
--- a/net/dsa/tag_dsa.c
+++ b/net/dsa/tag_dsa.c
@@ -246,12 +246,14 @@ static struct sk_buff *dsa_rcv_ll(struct sk_buff *skb, struct net_device *dev,
 
 	if (trunk) {
 		struct dsa_port *cpu_dp = dev->dsa_ptr;
+		struct dsa_lag *lag;
 
 		/* The exact source port is not available in the tag,
 		 * so we inject the frame directly on the upper
 		 * team/bond.
 		 */
-		skb->dev = dsa_lag_dev(cpu_dp->dst, source_port + 1);
+		lag = dsa_lag_by_id(cpu_dp->dst, source_port + 1);
+		skb->dev = lag ? lag->dev : NULL;
 	} else {
 		skb->dev = dsa_master_find_slave(dev, source_device,
 						 source_port);
-- 
2.25.1

Re: [PATCH v4 net-next 06/11] net: dsa: create a dsa_lag structure

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> The main purpose of this change is to create a data structure for a LAG
> as seen by DSA. This is similar to what we have for bridging - we pass a
> copy of this structure by value to ->port_lag_join and ->port_lag_leave.
> For now we keep the lag_dev, id and a reference count in it. Future
> patches will add a list of FDB entries for the LAG (these also need to
> be refcounted to work properly).
> 
> The LAG structure is created using dsa_port_lag_create() and destroyed
> using dsa_port_lag_destroy(), just like we have for bridging.
> 
> Because now, the dsa_lag itself is refcounted, we can simplify
> dsa_lag_map() and dsa_lag_unmap(). These functions need to keep a LAG in
> the dst->lags array only as long as at least one port uses it. The
> refcounting logic inside those functions can be removed now - they are
> called only when we should perform the operation.
> 
> dsa_lag_dev() is renamed to dsa_lag_by_id() and now returns the dsa_lag
> structure instead of the lag_dev net_device.
> 
> dsa_lag_foreach_port() now takes the dsa_lag structure as argument.
> 
> dst->lags holds an array of dsa_lag structures.
> 
> dsa_lag_map() now also saves the dsa_lag->id value, so that linear
> walking of dst->lags in drivers using dsa_lag_id() is no longer
> necessary. They can just look at lag.id.
> 
> dsa_port_lag_id_get() is a helper, similar to dsa_port_bridge_num_get(),
> which can be used by drivers to get the LAG ID assigned by DSA to a
> given port.
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 07/11] net: switchdev: remove lag_mod_cb from switchdev_handle_fdb_event_to_device

[Vladimir Oltean]

When the switchdev_handle_fdb_event_to_device() event replication helper
was created, my original thought was that FDB events on LAG interfaces
should most likely be special-cased, not just replicated towards all
switchdev ports beneath that LAG. So this replication helper currently
does not recurse through switchdev lower interfaces of LAG bridge ports,
but rather calls the lag_mod_cb() if that was provided.

No switchdev driver uses this helper for FDB events on LAG interfaces
yet, so that was an assumption which was yet to be tested. It is
certainly usable for that purpose, as my RFC series shows:

https://patchwork.kernel.org/project/netdevbpf/cover/20220210125201.2859463-1-vladimir.oltean@nxp.com/

however this approach is slightly convoluted because:

- the switchdev driver gets a "dev" that isn't its own net device, but
  rather the LAG net device. It must call switchdev_lower_dev_find(dev)
  in order to get a handle of any of its own net devices (the ones that
  pass check_cb).

- in order for FDB entries on LAG ports to be correctly refcounted per
  the number of switchdev ports beneath that LAG, we haven't escaped the
  need to iterate through the LAG's lower interfaces. Except that is now
  the responsibility of the switchdev driver, because the replication
  helper just stopped half-way.

So, even though yes, FDB events on LAG bridge ports must be
special-cased, in the end it's simpler to let switchdev_handle_fdb_*
just iterate through the LAG port's switchdev lowers, and let the
switchdev driver figure out that those physical ports are under a LAG.

The switchdev_handle_fdb_event_to_device() helper takes a
"foreign_dev_check" callback so it can figure out whether @dev can
autonomously forward to @foreign_dev. DSA fills this method properly:
if the LAG is offloaded by another port in the same tree as @dev, then
it isn't foreign. If it is a software LAG, it is foreign - forwarding
happens in software.

Whether an interface is foreign or not decides whether the replication
helper will go through the LAG's switchdev lowers or not. Since the
lan966x doesn't properly fill this out, FDB events on software LAG
uppers will get called. By changing lan966x_foreign_dev_check(), we can
suppress them.

Whereas DSA will now start receiving FDB events for its offloaded LAG
uppers, so we need to return -EOPNOTSUPP, since we currently don't do
the right thing for them.

Cc: Horatiu Vultur <horatiu.vultur@microchip.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v3->v4: none
v2->v3: patch is new, logically replaces previous patch "net: switchdev:
        export switchdev_lower_dev_find"

 .../microchip/lan966x/lan966x_switchdev.c     | 12 +--
 include/net/switchdev.h                       | 10 +--
 net/dsa/slave.c                               |  6 +-
 net/switchdev/switchdev.c                     | 80 +++++++------------
 4 files changed, 42 insertions(+), 66 deletions(-)

diff --git a/drivers/net/ethernet/microchip/lan966x/lan966x_switchdev.c b/drivers/net/ethernet/microchip/lan966x/lan966x_switchdev.c
index 85099a51d4c7..e3555c94294d 100644
--- a/drivers/net/ethernet/microchip/lan966x/lan966x_switchdev.c
+++ b/drivers/net/ethernet/microchip/lan966x/lan966x_switchdev.c
@@ -419,6 +419,9 @@ static int lan966x_netdevice_event(struct notifier_block *nb,
 	return notifier_from_errno(ret);
 }
 
+/* We don't offload uppers such as LAG as bridge ports, so every device except
+ * the bridge itself is foreign.
+ */
 static bool lan966x_foreign_dev_check(const struct net_device *dev,
 				      const struct net_device *foreign_dev)
 {
@@ -426,10 +429,10 @@ static bool lan966x_foreign_dev_check(const struct net_device *dev,
 	struct lan966x *lan966x = port->lan966x;
 
 	if (netif_is_bridge_master(foreign_dev))
-		if (lan966x->bridge != foreign_dev)
-			return true;
+		if (lan966x->bridge == foreign_dev)
+			return false;
 
-	return false;
+	return true;
 }
 
 static int lan966x_switchdev_event(struct notifier_block *nb,
@@ -449,8 +452,7 @@ static int lan966x_switchdev_event(struct notifier_block *nb,
 		err = switchdev_handle_fdb_event_to_device(dev, event, ptr,
 							   lan966x_netdevice_check,
 							   lan966x_foreign_dev_check,
-							   lan966x_handle_fdb,
-							   NULL);
+							   lan966x_handle_fdb);
 		return notifier_from_errno(err);
 	}
 
diff --git a/include/net/switchdev.h b/include/net/switchdev.h
index c32e1c8f79ec..3e424d40fae3 100644
--- a/include/net/switchdev.h
+++ b/include/net/switchdev.h
@@ -313,10 +313,7 @@ int switchdev_handle_fdb_event_to_device(struct net_device *dev, unsigned long e
 					     const struct net_device *foreign_dev),
 		int (*mod_cb)(struct net_device *dev, struct net_device *orig_dev,
 			      unsigned long event, const void *ctx,
-			      const struct switchdev_notifier_fdb_info *fdb_info),
-		int (*lag_mod_cb)(struct net_device *dev, struct net_device *orig_dev,
-				  unsigned long event, const void *ctx,
-				  const struct switchdev_notifier_fdb_info *fdb_info));
+			      const struct switchdev_notifier_fdb_info *fdb_info));
 
 int switchdev_handle_port_obj_add(struct net_device *dev,
 			struct switchdev_notifier_port_obj_info *port_obj_info,
@@ -443,10 +440,7 @@ switchdev_handle_fdb_event_to_device(struct net_device *dev, unsigned long event
 					     const struct net_device *foreign_dev),
 		int (*mod_cb)(struct net_device *dev, struct net_device *orig_dev,
 			      unsigned long event, const void *ctx,
-			      const struct switchdev_notifier_fdb_info *fdb_info),
-		int (*lag_mod_cb)(struct net_device *dev, struct net_device *orig_dev,
-				  unsigned long event, const void *ctx,
-				  const struct switchdev_notifier_fdb_info *fdb_info))
+			      const struct switchdev_notifier_fdb_info *fdb_info))
 {
 	return 0;
 }
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index e31c7710fee9..4ea6e0fd4b99 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -2461,6 +2461,9 @@ static int dsa_slave_fdb_event(struct net_device *dev,
 	bool host_addr = fdb_info->is_local;
 	struct dsa_switch *ds = dp->ds;
 
+	if (dp->lag)
+		return -EOPNOTSUPP;
+
 	if (ctx && ctx != dp)
 		return 0;
 
@@ -2526,8 +2529,7 @@ static int dsa_slave_switchdev_event(struct notifier_block *unused,
 		err = switchdev_handle_fdb_event_to_device(dev, event, ptr,
 							   dsa_slave_dev_check,
 							   dsa_foreign_dev_check,
-							   dsa_slave_fdb_event,
-							   NULL);
+							   dsa_slave_fdb_event);
 		return notifier_from_errno(err);
 	default:
 		return NOTIFY_DONE;
diff --git a/net/switchdev/switchdev.c b/net/switchdev/switchdev.c
index 28d2ccfe109c..474f76383033 100644
--- a/net/switchdev/switchdev.c
+++ b/net/switchdev/switchdev.c
@@ -458,63 +458,40 @@ static int __switchdev_handle_fdb_event_to_device(struct net_device *dev,
 					     const struct net_device *foreign_dev),
 		int (*mod_cb)(struct net_device *dev, struct net_device *orig_dev,
 			      unsigned long event, const void *ctx,
-			      const struct switchdev_notifier_fdb_info *fdb_info),
-		int (*lag_mod_cb)(struct net_device *dev, struct net_device *orig_dev,
-				  unsigned long event, const void *ctx,
-				  const struct switchdev_notifier_fdb_info *fdb_info))
+			      const struct switchdev_notifier_fdb_info *fdb_info))
 {
 	const struct switchdev_notifier_info *info = &fdb_info->info;
-	struct net_device *br, *lower_dev;
+	struct net_device *br, *lower_dev, *switchdev;
 	struct list_head *iter;
 	int err = -EOPNOTSUPP;
 
 	if (check_cb(dev))
 		return mod_cb(dev, orig_dev, event, info->ctx, fdb_info);
 
-	if (netif_is_lag_master(dev)) {
-		if (!switchdev_lower_dev_find_rcu(dev, check_cb, foreign_dev_check_cb))
-			goto maybe_bridged_with_us;
-
-		/* This is a LAG interface that we offload */
-		if (!lag_mod_cb)
-			return -EOPNOTSUPP;
-
-		return lag_mod_cb(dev, orig_dev, event, info->ctx, fdb_info);
-	}
-
 	/* Recurse through lower interfaces in case the FDB entry is pointing
-	 * towards a bridge device.
+	 * towards a bridge or a LAG device.
 	 */
-	if (netif_is_bridge_master(dev)) {
-		if (!switchdev_lower_dev_find_rcu(dev, check_cb, foreign_dev_check_cb))
-			return 0;
-
-		/* This is a bridge interface that we offload */
-		netdev_for_each_lower_dev(dev, lower_dev, iter) {
-			/* Do not propagate FDB entries across bridges */
-			if (netif_is_bridge_master(lower_dev))
-				continue;
-
-			/* Bridge ports might be either us, or LAG interfaces
-			 * that we offload.
-			 */
-			if (!check_cb(lower_dev) &&
-			    !switchdev_lower_dev_find_rcu(lower_dev, check_cb,
-							  foreign_dev_check_cb))
-				continue;
-
-			err = __switchdev_handle_fdb_event_to_device(lower_dev, orig_dev,
-								     event, fdb_info, check_cb,
-								     foreign_dev_check_cb,
-								     mod_cb, lag_mod_cb);
-			if (err && err != -EOPNOTSUPP)
-				return err;
-		}
+	netdev_for_each_lower_dev(dev, lower_dev, iter) {
+		/* Do not propagate FDB entries across bridges */
+		if (netif_is_bridge_master(lower_dev))
+			continue;
 
-		return 0;
+		/* Bridge ports might be either us, or LAG interfaces
+		 * that we offload.
+		 */
+		if (!check_cb(lower_dev) &&
+		    !switchdev_lower_dev_find_rcu(lower_dev, check_cb,
+						  foreign_dev_check_cb))
+			continue;
+
+		err = __switchdev_handle_fdb_event_to_device(lower_dev, orig_dev,
+							     event, fdb_info, check_cb,
+							     foreign_dev_check_cb,
+							     mod_cb);
+		if (err && err != -EOPNOTSUPP)
+			return err;
 	}
 
-maybe_bridged_with_us:
 	/* Event is neither on a bridge nor a LAG. Check whether it is on an
 	 * interface that is in a bridge with us.
 	 */
@@ -522,12 +499,16 @@ static int __switchdev_handle_fdb_event_to_device(struct net_device *dev,
 	if (!br || !netif_is_bridge_master(br))
 		return 0;
 
-	if (!switchdev_lower_dev_find_rcu(br, check_cb, foreign_dev_check_cb))
+	switchdev = switchdev_lower_dev_find_rcu(br, check_cb, foreign_dev_check_cb);
+	if (!switchdev)
 		return 0;
 
+	if (!foreign_dev_check_cb(switchdev, dev))
+		return err;
+
 	return __switchdev_handle_fdb_event_to_device(br, orig_dev, event, fdb_info,
 						      check_cb, foreign_dev_check_cb,
-						      mod_cb, lag_mod_cb);
+						      mod_cb);
 }
 
 int switchdev_handle_fdb_event_to_device(struct net_device *dev, unsigned long event,
@@ -537,16 +518,13 @@ int switchdev_handle_fdb_event_to_device(struct net_device *dev, unsigned long e
 					     const struct net_device *foreign_dev),
 		int (*mod_cb)(struct net_device *dev, struct net_device *orig_dev,
 			      unsigned long event, const void *ctx,
-			      const struct switchdev_notifier_fdb_info *fdb_info),
-		int (*lag_mod_cb)(struct net_device *dev, struct net_device *orig_dev,
-				  unsigned long event, const void *ctx,
-				  const struct switchdev_notifier_fdb_info *fdb_info))
+			      const struct switchdev_notifier_fdb_info *fdb_info))
 {
 	int err;
 
 	err = __switchdev_handle_fdb_event_to_device(dev, dev, event, fdb_info,
 						     check_cb, foreign_dev_check_cb,
-						     mod_cb, lag_mod_cb);
+						     mod_cb);
 	if (err == -EOPNOTSUPP)
 		err = 0;
 
-- 
2.25.1

Re: [PATCH v4 net-next 07/11] net: switchdev: remove lag_mod_cb from switchdev_handle_fdb_event_to_device

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> When the switchdev_handle_fdb_event_to_device() event replication helper
> was created, my original thought was that FDB events on LAG interfaces
> should most likely be special-cased, not just replicated towards all
> switchdev ports beneath that LAG. So this replication helper currently
> does not recurse through switchdev lower interfaces of LAG bridge ports,
> but rather calls the lag_mod_cb() if that was provided.
> 
> No switchdev driver uses this helper for FDB events on LAG interfaces
> yet, so that was an assumption which was yet to be tested. It is
> certainly usable for that purpose, as my RFC series shows:
> 
> https://patchwork.kernel.org/project/netdevbpf/cover/20220210125201.2859463-1-vladimir.oltean@nxp.com/
> 
> however this approach is slightly convoluted because:
> 
> - the switchdev driver gets a "dev" that isn't its own net device, but
>    rather the LAG net device. It must call switchdev_lower_dev_find(dev)
>    in order to get a handle of any of its own net devices (the ones that
>    pass check_cb).
> 
> - in order for FDB entries on LAG ports to be correctly refcounted per
>    the number of switchdev ports beneath that LAG, we haven't escaped the
>    need to iterate through the LAG's lower interfaces. Except that is now
>    the responsibility of the switchdev driver, because the replication
>    helper just stopped half-way.
> 
> So, even though yes, FDB events on LAG bridge ports must be
> special-cased, in the end it's simpler to let switchdev_handle_fdb_*
> just iterate through the LAG port's switchdev lowers, and let the
> switchdev driver figure out that those physical ports are under a LAG.
> 
> The switchdev_handle_fdb_event_to_device() helper takes a
> "foreign_dev_check" callback so it can figure out whether @dev can
> autonomously forward to @foreign_dev. DSA fills this method properly:
> if the LAG is offloaded by another port in the same tree as @dev, then
> it isn't foreign. If it is a software LAG, it is foreign - forwarding
> happens in software.
> 
> Whether an interface is foreign or not decides whether the replication
> helper will go through the LAG's switchdev lowers or not. Since the
> lan966x doesn't properly fill this out, FDB events on software LAG
> uppers will get called. By changing lan966x_foreign_dev_check(), we can
> suppress them.
> 
> Whereas DSA will now start receiving FDB events for its offloaded LAG
> uppers, so we need to return -EOPNOTSUPP, since we currently don't do
> the right thing for them.
> 
> Cc: Horatiu Vultur <horatiu.vultur@microchip.com>
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 08/11] net: dsa: remove "ds" and "port" from struct dsa_switchdev_event_work

[Vladimir Oltean]

By construction, the struct net_device *dev passed to
dsa_slave_switchdev_event_work() via struct dsa_switchdev_event_work
is always a DSA slave device.

Therefore, it is redundant to pass struct dsa_switch and int port
information in the deferred work structure. This can be retrieved at all
times from the provided struct net_device via dsa_slave_to_port().

For the same reason, we can drop the dsa_is_user_port() check in
dsa_fdb_offload_notify().

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v4: none

 net/dsa/dsa_priv.h |  2 --
 net/dsa/slave.c    | 16 +++++-----------
 2 files changed, 5 insertions(+), 13 deletions(-)

diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
index 8612ff8ea7fe..f35b7a1496e1 100644
--- a/net/dsa/dsa_priv.h
+++ b/net/dsa/dsa_priv.h
@@ -119,8 +119,6 @@ struct dsa_notifier_master_state_info {
 };
 
 struct dsa_switchdev_event_work {
-	struct dsa_switch *ds;
-	int port;
 	struct net_device *dev;
 	struct work_struct work;
 	unsigned long event;
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index 4ea6e0fd4b99..7eb972691ce9 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -2373,29 +2373,25 @@ static void
 dsa_fdb_offload_notify(struct dsa_switchdev_event_work *switchdev_work)
 {
 	struct switchdev_notifier_fdb_info info = {};
-	struct dsa_switch *ds = switchdev_work->ds;
-	struct dsa_port *dp;
-
-	if (!dsa_is_user_port(ds, switchdev_work->port))
-		return;
 
 	info.addr = switchdev_work->addr;
 	info.vid = switchdev_work->vid;
 	info.offloaded = true;
-	dp = dsa_to_port(ds, switchdev_work->port);
 	call_switchdev_notifiers(SWITCHDEV_FDB_OFFLOADED,
-				 dp->slave, &info.info, NULL);
+				 switchdev_work->dev, &info.info, NULL);
 }
 
 static void dsa_slave_switchdev_event_work(struct work_struct *work)
 {
 	struct dsa_switchdev_event_work *switchdev_work =
 		container_of(work, struct dsa_switchdev_event_work, work);
-	struct dsa_switch *ds = switchdev_work->ds;
+	struct net_device *dev = switchdev_work->dev;
+	struct dsa_switch *ds;
 	struct dsa_port *dp;
 	int err;
 
-	dp = dsa_to_port(ds, switchdev_work->port);
+	dp = dsa_slave_to_port(dev);
+	ds = dp->ds;
 
 	switch (switchdev_work->event) {
 	case SWITCHDEV_FDB_ADD_TO_DEVICE:
@@ -2497,8 +2493,6 @@ static int dsa_slave_fdb_event(struct net_device *dev,
 		   host_addr ? " as host address" : "");
 
 	INIT_WORK(&switchdev_work->work, dsa_slave_switchdev_event_work);
-	switchdev_work->ds = ds;
-	switchdev_work->port = dp->index;
 	switchdev_work->event = event;
 	switchdev_work->dev = dev;
 
-- 
2.25.1

Re: [PATCH v4 net-next 08/11] net: dsa: remove "ds" and "port" from struct dsa_switchdev_event_work

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> By construction, the struct net_device *dev passed to
> dsa_slave_switchdev_event_work() via struct dsa_switchdev_event_work
> is always a DSA slave device.
> 
> Therefore, it is redundant to pass struct dsa_switch and int port
> information in the deferred work structure. This can be retrieved at all
> times from the provided struct net_device via dsa_slave_to_port().
> 
> For the same reason, we can drop the dsa_is_user_port() check in
> dsa_fdb_offload_notify().
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 09/11] net: dsa: call SWITCHDEV_FDB_OFFLOADED for the orig_dev

[Vladimir Oltean]

When switchdev_handle_fdb_event_to_device() replicates a FDB event
emitted for the bridge or for a LAG port and DSA offloads that, we
should notify back to switchdev that the FDB entry on the original
device is what was offloaded, not on the DSA slave devices that the
event is replicated on.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v2->v4: none

 net/dsa/dsa_priv.h | 1 +
 net/dsa/slave.c    | 3 ++-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
index f35b7a1496e1..f2364c5adc04 100644
--- a/net/dsa/dsa_priv.h
+++ b/net/dsa/dsa_priv.h
@@ -120,6 +120,7 @@ struct dsa_notifier_master_state_info {
 
 struct dsa_switchdev_event_work {
 	struct net_device *dev;
+	struct net_device *orig_dev;
 	struct work_struct work;
 	unsigned long event;
 	/* Specific for SWITCHDEV_FDB_ADD_TO_DEVICE and
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index 7eb972691ce9..4aeb3e092dd6 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -2378,7 +2378,7 @@ dsa_fdb_offload_notify(struct dsa_switchdev_event_work *switchdev_work)
 	info.vid = switchdev_work->vid;
 	info.offloaded = true;
 	call_switchdev_notifiers(SWITCHDEV_FDB_OFFLOADED,
-				 switchdev_work->dev, &info.info, NULL);
+				 switchdev_work->orig_dev, &info.info, NULL);
 }
 
 static void dsa_slave_switchdev_event_work(struct work_struct *work)
@@ -2495,6 +2495,7 @@ static int dsa_slave_fdb_event(struct net_device *dev,
 	INIT_WORK(&switchdev_work->work, dsa_slave_switchdev_event_work);
 	switchdev_work->event = event;
 	switchdev_work->dev = dev;
+	switchdev_work->orig_dev = orig_dev;
 
 	ether_addr_copy(switchdev_work->addr, fdb_info->addr);
 	switchdev_work->vid = fdb_info->vid;
-- 
2.25.1

Re: [PATCH v4 net-next 09/11] net: dsa: call SWITCHDEV_FDB_OFFLOADED for the orig_dev

[Florian Fainelli]

On 2/21/2022 1:23 PM, Vladimir Oltean wrote:
> When switchdev_handle_fdb_event_to_device() replicates a FDB event
> emitted for the bridge or for a LAG port and DSA offloads that, we
> should notify back to switchdev that the FDB entry on the original
> device is what was offloaded, not on the DSA slave devices that the
> event is replicated on.
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

[PATCH v4 net-next 10/11] net: dsa: support FDB events on offloaded LAG interfaces

[Vladimir Oltean]

This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.

Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.

To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.

To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
  while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
  we don't want FDB entries lingering around, but we don't want to
  remove entries that are still in use, either

For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
  the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
  by one
- if an FDB is installed towards a LAG with ports already present, that
  entry is created (if it doesn't exist) and its refcount is bumped by
  the amount of ports already present in the LAG

echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static

ip link del br0
ip link del bond0

echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up

ip link del br0
ip link del bond0

echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up

ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v3->v4:
- remove the "void *ctx" left over in struct dsa_switchdev_event_work
- make sure the dp->lag assignment is last in dsa_port_lag_create()
v2->v3:
- leave iteration among DSA slave interfaces that are members of
  the LAG bridge port to switchdev_handle_fdb_event_to_device()
- reorder some checks that previously resulted in the access of an
  uninitialized "ds" pointer

 include/net/dsa.h  |   6 +++
 net/dsa/dsa_priv.h |  13 ++++++
 net/dsa/port.c     |  27 +++++++++++
 net/dsa/slave.c    |  43 +++++++++++-------
 net/dsa/switch.c   | 109 +++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 183 insertions(+), 15 deletions(-)

diff --git a/include/net/dsa.h b/include/net/dsa.h
index 81ed34998416..01faba89c987 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -119,6 +119,8 @@ struct dsa_netdevice_ops {
 struct dsa_lag {
 	struct net_device *dev;
 	unsigned int id;
+	struct mutex fdb_lock;
+	struct list_head fdbs;
 	refcount_t refcount;
 };
 
@@ -944,6 +946,10 @@ struct dsa_switch_ops {
 				const unsigned char *addr, u16 vid);
 	int	(*port_fdb_dump)(struct dsa_switch *ds, int port,
 				 dsa_fdb_dump_cb_t *cb, void *data);
+	int	(*lag_fdb_add)(struct dsa_switch *ds, struct dsa_lag lag,
+			       const unsigned char *addr, u16 vid);
+	int	(*lag_fdb_del)(struct dsa_switch *ds, struct dsa_lag lag,
+			       const unsigned char *addr, u16 vid);
 
 	/*
 	 * Multicast database
diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
index f2364c5adc04..1ba93afdc874 100644
--- a/net/dsa/dsa_priv.h
+++ b/net/dsa/dsa_priv.h
@@ -25,6 +25,8 @@ enum {
 	DSA_NOTIFIER_FDB_DEL,
 	DSA_NOTIFIER_HOST_FDB_ADD,
 	DSA_NOTIFIER_HOST_FDB_DEL,
+	DSA_NOTIFIER_LAG_FDB_ADD,
+	DSA_NOTIFIER_LAG_FDB_DEL,
 	DSA_NOTIFIER_LAG_CHANGE,
 	DSA_NOTIFIER_LAG_JOIN,
 	DSA_NOTIFIER_LAG_LEAVE,
@@ -67,6 +69,13 @@ struct dsa_notifier_fdb_info {
 	u16 vid;
 };
 
+/* DSA_NOTIFIER_LAG_FDB_* */
+struct dsa_notifier_lag_fdb_info {
+	struct dsa_lag *lag;
+	const unsigned char *addr;
+	u16 vid;
+};
+
 /* DSA_NOTIFIER_MDB_* */
 struct dsa_notifier_mdb_info {
 	const struct switchdev_obj_port_mdb *mdb;
@@ -214,6 +223,10 @@ int dsa_port_host_fdb_add(struct dsa_port *dp, const unsigned char *addr,
 			  u16 vid);
 int dsa_port_host_fdb_del(struct dsa_port *dp, const unsigned char *addr,
 			  u16 vid);
+int dsa_port_lag_fdb_add(struct dsa_port *dp, const unsigned char *addr,
+			 u16 vid);
+int dsa_port_lag_fdb_del(struct dsa_port *dp, const unsigned char *addr,
+			 u16 vid);
 int dsa_port_fdb_dump(struct dsa_port *dp, dsa_fdb_dump_cb_t *cb, void *data);
 int dsa_port_mdb_add(const struct dsa_port *dp,
 		     const struct switchdev_obj_port_mdb *mdb);
diff --git a/net/dsa/port.c b/net/dsa/port.c
index bf30ceaf39f4..9ca9673c9198 100644
--- a/net/dsa/port.c
+++ b/net/dsa/port.c
@@ -465,6 +465,8 @@ static int dsa_port_lag_create(struct dsa_port *dp,
 		return -ENOMEM;
 
 	refcount_set(&lag->refcount, 1);
+	mutex_init(&lag->fdb_lock);
+	INIT_LIST_HEAD(&lag->fdbs);
 	lag->dev = lag_dev;
 	dsa_lag_map(ds->dst, lag);
 	dp->lag = lag;
@@ -482,6 +484,7 @@ static void dsa_port_lag_destroy(struct dsa_port *dp)
 	if (!refcount_dec_and_test(&lag->refcount))
 		return;
 
+	WARN_ON(!list_empty(&lag->fdbs));
 	dsa_lag_unmap(dp->ds->dst, lag);
 	kfree(lag);
 }
@@ -860,6 +863,30 @@ int dsa_port_host_fdb_del(struct dsa_port *dp, const unsigned char *addr,
 	return dsa_port_notify(dp, DSA_NOTIFIER_HOST_FDB_DEL, &info);
 }
 
+int dsa_port_lag_fdb_add(struct dsa_port *dp, const unsigned char *addr,
+			 u16 vid)
+{
+	struct dsa_notifier_lag_fdb_info info = {
+		.lag = dp->lag,
+		.addr = addr,
+		.vid = vid,
+	};
+
+	return dsa_port_notify(dp, DSA_NOTIFIER_LAG_FDB_ADD, &info);
+}
+
+int dsa_port_lag_fdb_del(struct dsa_port *dp, const unsigned char *addr,
+			 u16 vid)
+{
+	struct dsa_notifier_lag_fdb_info info = {
+		.lag = dp->lag,
+		.addr = addr,
+		.vid = vid,
+	};
+
+	return dsa_port_notify(dp, DSA_NOTIFIER_LAG_FDB_DEL, &info);
+}
+
 int dsa_port_fdb_dump(struct dsa_port *dp, dsa_fdb_dump_cb_t *cb, void *data)
 {
 	struct dsa_switch *ds = dp->ds;
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index 4aeb3e092dd6..089616206b11 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -2398,6 +2398,9 @@ static void dsa_slave_switchdev_event_work(struct work_struct *work)
 		if (switchdev_work->host_addr)
 			err = dsa_port_host_fdb_add(dp, switchdev_work->addr,
 						    switchdev_work->vid);
+		else if (dp->lag)
+			err = dsa_port_lag_fdb_add(dp, switchdev_work->addr,
+						   switchdev_work->vid);
 		else
 			err = dsa_port_fdb_add(dp, switchdev_work->addr,
 					       switchdev_work->vid);
@@ -2415,6 +2418,9 @@ static void dsa_slave_switchdev_event_work(struct work_struct *work)
 		if (switchdev_work->host_addr)
 			err = dsa_port_host_fdb_del(dp, switchdev_work->addr,
 						    switchdev_work->vid);
+		else if (dp->lag)
+			err = dsa_port_lag_fdb_del(dp, switchdev_work->addr,
+						   switchdev_work->vid);
 		else
 			err = dsa_port_fdb_del(dp, switchdev_work->addr,
 					       switchdev_work->vid);
@@ -2457,25 +2463,20 @@ static int dsa_slave_fdb_event(struct net_device *dev,
 	bool host_addr = fdb_info->is_local;
 	struct dsa_switch *ds = dp->ds;
 
-	if (dp->lag)
-		return -EOPNOTSUPP;
-
 	if (ctx && ctx != dp)
 		return 0;
 
-	if (!ds->ops->port_fdb_add || !ds->ops->port_fdb_del)
-		return -EOPNOTSUPP;
-
-	if (dsa_slave_dev_check(orig_dev) &&
-	    switchdev_fdb_is_dynamically_learned(fdb_info))
-		return 0;
+	if (switchdev_fdb_is_dynamically_learned(fdb_info)) {
+		if (dsa_port_offloads_bridge_port(dp, orig_dev))
+			return 0;
 
-	/* FDB entries learned by the software bridge should be installed as
-	 * host addresses only if the driver requests assisted learning.
-	 */
-	if (switchdev_fdb_is_dynamically_learned(fdb_info) &&
-	    !ds->assisted_learning_on_cpu_port)
-		return 0;
+		/* FDB entries learned by the software bridge or by foreign
+		 * bridge ports should be installed as host addresses only if
+		 * the driver requests assisted learning.
+		 */
+		if (!ds->assisted_learning_on_cpu_port)
+			return 0;
+	}
 
 	/* Also treat FDB entries on foreign interfaces bridged with us as host
 	 * addresses.
@@ -2483,6 +2484,18 @@ static int dsa_slave_fdb_event(struct net_device *dev,
 	if (dsa_foreign_dev_check(dev, orig_dev))
 		host_addr = true;
 
+	/* Check early that we're not doing work in vain.
+	 * Host addresses on LAG ports still require regular FDB ops,
+	 * since the CPU port isn't in a LAG.
+	 */
+	if (dp->lag && !host_addr) {
+		if (!ds->ops->lag_fdb_add || !ds->ops->lag_fdb_del)
+			return -EOPNOTSUPP;
+	} else {
+		if (!ds->ops->port_fdb_add || !ds->ops->port_fdb_del)
+			return -EOPNOTSUPP;
+	}
+
 	switchdev_work = kzalloc(sizeof(*switchdev_work), GFP_ATOMIC);
 	if (!switchdev_work)
 		return -ENOMEM;
diff --git a/net/dsa/switch.c b/net/dsa/switch.c
index 0bb3987bd4e6..0c2961cbc105 100644
--- a/net/dsa/switch.c
+++ b/net/dsa/switch.c
@@ -385,6 +385,75 @@ static int dsa_port_do_fdb_del(struct dsa_port *dp, const unsigned char *addr,
 	return err;
 }
 
+static int dsa_switch_do_lag_fdb_add(struct dsa_switch *ds, struct dsa_lag *lag,
+				     const unsigned char *addr, u16 vid)
+{
+	struct dsa_mac_addr *a;
+	int err = 0;
+
+	mutex_lock(&lag->fdb_lock);
+
+	a = dsa_mac_addr_find(&lag->fdbs, addr, vid);
+	if (a) {
+		refcount_inc(&a->refcount);
+		goto out;
+	}
+
+	a = kzalloc(sizeof(*a), GFP_KERNEL);
+	if (!a) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	err = ds->ops->lag_fdb_add(ds, *lag, addr, vid);
+	if (err) {
+		kfree(a);
+		goto out;
+	}
+
+	ether_addr_copy(a->addr, addr);
+	a->vid = vid;
+	refcount_set(&a->refcount, 1);
+	list_add_tail(&a->list, &lag->fdbs);
+
+out:
+	mutex_unlock(&lag->fdb_lock);
+
+	return err;
+}
+
+static int dsa_switch_do_lag_fdb_del(struct dsa_switch *ds, struct dsa_lag *lag,
+				     const unsigned char *addr, u16 vid)
+{
+	struct dsa_mac_addr *a;
+	int err = 0;
+
+	mutex_lock(&lag->fdb_lock);
+
+	a = dsa_mac_addr_find(&lag->fdbs, addr, vid);
+	if (!a) {
+		err = -ENOENT;
+		goto out;
+	}
+
+	if (!refcount_dec_and_test(&a->refcount))
+		goto out;
+
+	err = ds->ops->lag_fdb_del(ds, *lag, addr, vid);
+	if (err) {
+		refcount_set(&a->refcount, 1);
+		goto out;
+	}
+
+	list_del(&a->list);
+	kfree(a);
+
+out:
+	mutex_unlock(&lag->fdb_lock);
+
+	return err;
+}
+
 static int dsa_switch_host_fdb_add(struct dsa_switch *ds,
 				   struct dsa_notifier_fdb_info *info)
 {
@@ -451,6 +520,40 @@ static int dsa_switch_fdb_del(struct dsa_switch *ds,
 	return dsa_port_do_fdb_del(dp, info->addr, info->vid);
 }
 
+static int dsa_switch_lag_fdb_add(struct dsa_switch *ds,
+				  struct dsa_notifier_lag_fdb_info *info)
+{
+	struct dsa_port *dp;
+
+	if (!ds->ops->lag_fdb_add)
+		return -EOPNOTSUPP;
+
+	/* Notify switch only if it has a port in this LAG */
+	dsa_switch_for_each_port(dp, ds)
+		if (dsa_port_offloads_lag(dp, info->lag))
+			return dsa_switch_do_lag_fdb_add(ds, info->lag,
+							 info->addr, info->vid);
+
+	return 0;
+}
+
+static int dsa_switch_lag_fdb_del(struct dsa_switch *ds,
+				  struct dsa_notifier_lag_fdb_info *info)
+{
+	struct dsa_port *dp;
+
+	if (!ds->ops->lag_fdb_del)
+		return -EOPNOTSUPP;
+
+	/* Notify switch only if it has a port in this LAG */
+	dsa_switch_for_each_port(dp, ds)
+		if (dsa_port_offloads_lag(dp, info->lag))
+			return dsa_switch_do_lag_fdb_del(ds, info->lag,
+							 info->addr, info->vid);
+
+	return 0;
+}
+
 static int dsa_switch_lag_change(struct dsa_switch *ds,
 				 struct dsa_notifier_lag_info *info)
 {
@@ -904,6 +1007,12 @@ static int dsa_switch_event(struct notifier_block *nb,
 	case DSA_NOTIFIER_HOST_FDB_DEL:
 		err = dsa_switch_host_fdb_del(ds, info);
 		break;
+	case DSA_NOTIFIER_LAG_FDB_ADD:
+		err = dsa_switch_lag_fdb_add(ds, info);
+		break;
+	case DSA_NOTIFIER_LAG_FDB_DEL:
+		err = dsa_switch_lag_fdb_del(ds, info);
+		break;
 	case DSA_NOTIFIER_LAG_CHANGE:
 		err = dsa_switch_lag_change(ds, info);
 		break;
-- 
2.25.1

[PATCH v4 net-next 11/11] net: dsa: felix: support FDB entries on offloaded LAG interfaces

[Vladimir Oltean]

This adds the logic in the Felix DSA driver and Ocelot switch library.
For Ocelot switches, the DEST_IDX that is the output of the MAC table
lookup is a logical port (equal to physical port, if no LAG is used, or
a dynamically allocated number otherwise). The allocation we have in
place for LAG IDs is different from DSA's, so we can't use that:
- DSA allocates a continuous range of LAG IDs starting from 1
- Ocelot appears to require that physical ports and LAG IDs are in the
  same space of [0, num_phys_ports), and additionally, ports that aren't
  in a LAG must have physical port id == logical port id

The implication is that an FDB entry towards a LAG might need to be
deleted and reinstalled when the LAG ID changes.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v4: none

 drivers/net/dsa/ocelot/felix.c     |  18 ++++
 drivers/net/ethernet/mscc/ocelot.c | 128 ++++++++++++++++++++++++++++-
 include/soc/mscc/ocelot.h          |  12 +++
 3 files changed, 157 insertions(+), 1 deletion(-)

diff --git a/drivers/net/dsa/ocelot/felix.c b/drivers/net/dsa/ocelot/felix.c
index 6d483887af04..9959407fede8 100644
--- a/drivers/net/dsa/ocelot/felix.c
+++ b/drivers/net/dsa/ocelot/felix.c
@@ -614,6 +614,22 @@ static int felix_fdb_del(struct dsa_switch *ds, int port,
 	return ocelot_fdb_del(ocelot, port, addr, vid);
 }
 
+static int felix_lag_fdb_add(struct dsa_switch *ds, struct dsa_lag lag,
+			     const unsigned char *addr, u16 vid)
+{
+	struct ocelot *ocelot = ds->priv;
+
+	return ocelot_lag_fdb_add(ocelot, lag.dev, addr, vid);
+}
+
+static int felix_lag_fdb_del(struct dsa_switch *ds, struct dsa_lag lag,
+			     const unsigned char *addr, u16 vid)
+{
+	struct ocelot *ocelot = ds->priv;
+
+	return ocelot_lag_fdb_del(ocelot, lag.dev, addr, vid);
+}
+
 static int felix_mdb_add(struct dsa_switch *ds, int port,
 			 const struct switchdev_obj_port_mdb *mdb)
 {
@@ -1579,6 +1595,8 @@ const struct dsa_switch_ops felix_switch_ops = {
 	.port_fdb_dump			= felix_fdb_dump,
 	.port_fdb_add			= felix_fdb_add,
 	.port_fdb_del			= felix_fdb_del,
+	.lag_fdb_add			= felix_lag_fdb_add,
+	.lag_fdb_del			= felix_lag_fdb_del,
 	.port_mdb_add			= felix_mdb_add,
 	.port_mdb_del			= felix_mdb_del,
 	.port_pre_bridge_flags		= felix_pre_bridge_flags,
diff --git a/drivers/net/ethernet/mscc/ocelot.c b/drivers/net/ethernet/mscc/ocelot.c
index 2fb713e9baa4..0e8fa0a4fc69 100644
--- a/drivers/net/ethernet/mscc/ocelot.c
+++ b/drivers/net/ethernet/mscc/ocelot.c
@@ -1907,6 +1907,8 @@ static u32 ocelot_get_bond_mask(struct ocelot *ocelot, struct net_device *bond)
 	u32 mask = 0;
 	int port;
 
+	lockdep_assert_held(&ocelot->fwd_domain_lock);
+
 	for (port = 0; port < ocelot->num_phys_ports; port++) {
 		struct ocelot_port *ocelot_port = ocelot->ports[port];
 
@@ -1920,6 +1922,19 @@ static u32 ocelot_get_bond_mask(struct ocelot *ocelot, struct net_device *bond)
 	return mask;
 }
 
+/* The logical port number of a LAG is equal to the lowest numbered physical
+ * port ID present in that LAG. It may change if that port ever leaves the LAG.
+ */
+static int ocelot_bond_get_id(struct ocelot *ocelot, struct net_device *bond)
+{
+	int bond_mask = ocelot_get_bond_mask(ocelot, bond);
+
+	if (!bond_mask)
+		return -ENOENT;
+
+	return __ffs(bond_mask);
+}
+
 u32 ocelot_get_bridge_fwd_mask(struct ocelot *ocelot, int src_port)
 {
 	struct ocelot_port *ocelot_port = ocelot->ports[src_port];
@@ -2413,7 +2428,7 @@ static void ocelot_setup_logical_port_ids(struct ocelot *ocelot)
 
 		bond = ocelot_port->bond;
 		if (bond) {
-			int lag = __ffs(ocelot_get_bond_mask(ocelot, bond));
+			int lag = ocelot_bond_get_id(ocelot, bond);
 
 			ocelot_rmw_gix(ocelot,
 				       ANA_PORT_PORT_CFG_PORTID_VAL(lag),
@@ -2428,6 +2443,46 @@ static void ocelot_setup_logical_port_ids(struct ocelot *ocelot)
 	}
 }
 
+/* Documentation for PORTID_VAL says:
+ *     Logical port number for front port. If port is not a member of a LLAG,
+ *     then PORTID must be set to the physical port number.
+ *     If port is a member of a LLAG, then PORTID must be set to the common
+ *     PORTID_VAL used for all member ports of the LLAG.
+ *     The value must not exceed the number of physical ports on the device.
+ *
+ * This means we have little choice but to migrate FDB entries pointing towards
+ * a logical port when that changes.
+ */
+static void ocelot_migrate_lag_fdbs(struct ocelot *ocelot,
+				    struct net_device *bond,
+				    int lag)
+{
+	struct ocelot_lag_fdb *fdb;
+	int err;
+
+	lockdep_assert_held(&ocelot->fwd_domain_lock);
+
+	list_for_each_entry(fdb, &ocelot->lag_fdbs, list) {
+		if (fdb->bond != bond)
+			continue;
+
+		err = ocelot_mact_forget(ocelot, fdb->addr, fdb->vid);
+		if (err) {
+			dev_err(ocelot->dev,
+				"failed to delete LAG %s FDB %pM vid %d: %pe\n",
+				bond->name, fdb->addr, fdb->vid, ERR_PTR(err));
+		}
+
+		err = ocelot_mact_learn(ocelot, lag, fdb->addr, fdb->vid,
+					ENTRYTYPE_LOCKED);
+		if (err) {
+			dev_err(ocelot->dev,
+				"failed to migrate LAG %s FDB %pM vid %d: %pe\n",
+				bond->name, fdb->addr, fdb->vid, ERR_PTR(err));
+		}
+	}
+}
+
 int ocelot_port_lag_join(struct ocelot *ocelot, int port,
 			 struct net_device *bond,
 			 struct netdev_lag_upper_info *info)
@@ -2452,14 +2507,23 @@ EXPORT_SYMBOL(ocelot_port_lag_join);
 void ocelot_port_lag_leave(struct ocelot *ocelot, int port,
 			   struct net_device *bond)
 {
+	int old_lag_id, new_lag_id;
+
 	mutex_lock(&ocelot->fwd_domain_lock);
 
+	old_lag_id = ocelot_bond_get_id(ocelot, bond);
+
 	ocelot->ports[port]->bond = NULL;
 
 	ocelot_setup_logical_port_ids(ocelot);
 	ocelot_apply_bridge_fwd_mask(ocelot, false);
 	ocelot_set_aggr_pgids(ocelot);
 
+	new_lag_id = ocelot_bond_get_id(ocelot, bond);
+
+	if (new_lag_id >= 0 && old_lag_id != new_lag_id)
+		ocelot_migrate_lag_fdbs(ocelot, bond, new_lag_id);
+
 	mutex_unlock(&ocelot->fwd_domain_lock);
 }
 EXPORT_SYMBOL(ocelot_port_lag_leave);
@@ -2468,13 +2532,74 @@ void ocelot_port_lag_change(struct ocelot *ocelot, int port, bool lag_tx_active)
 {
 	struct ocelot_port *ocelot_port = ocelot->ports[port];
 
+	mutex_lock(&ocelot->fwd_domain_lock);
+
 	ocelot_port->lag_tx_active = lag_tx_active;
 
 	/* Rebalance the LAGs */
 	ocelot_set_aggr_pgids(ocelot);
+
+	mutex_unlock(&ocelot->fwd_domain_lock);
 }
 EXPORT_SYMBOL(ocelot_port_lag_change);
 
+int ocelot_lag_fdb_add(struct ocelot *ocelot, struct net_device *bond,
+		       const unsigned char *addr, u16 vid)
+{
+	struct ocelot_lag_fdb *fdb;
+	int lag, err;
+
+	fdb = kzalloc(sizeof(*fdb), GFP_KERNEL);
+	if (!fdb)
+		return -ENOMEM;
+
+	ether_addr_copy(fdb->addr, addr);
+	fdb->vid = vid;
+	fdb->bond = bond;
+
+	mutex_lock(&ocelot->fwd_domain_lock);
+	lag = ocelot_bond_get_id(ocelot, bond);
+
+	err = ocelot_mact_learn(ocelot, lag, addr, vid, ENTRYTYPE_LOCKED);
+	if (err) {
+		mutex_unlock(&ocelot->fwd_domain_lock);
+		kfree(fdb);
+		return err;
+	}
+
+	list_add_tail(&fdb->list, &ocelot->lag_fdbs);
+	mutex_unlock(&ocelot->fwd_domain_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(ocelot_lag_fdb_add);
+
+int ocelot_lag_fdb_del(struct ocelot *ocelot, struct net_device *bond,
+		       const unsigned char *addr, u16 vid)
+{
+	struct ocelot_lag_fdb *fdb, *tmp;
+
+	mutex_lock(&ocelot->fwd_domain_lock);
+
+	list_for_each_entry_safe(fdb, tmp, &ocelot->lag_fdbs, list) {
+		if (!ether_addr_equal(fdb->addr, addr) || fdb->vid != vid ||
+		    fdb->bond != bond)
+			continue;
+
+		ocelot_mact_forget(ocelot, addr, vid);
+		list_del(&fdb->list);
+		mutex_unlock(&ocelot->fwd_domain_lock);
+		kfree(fdb);
+
+		return 0;
+	}
+
+	mutex_unlock(&ocelot->fwd_domain_lock);
+
+	return -ENOENT;
+}
+EXPORT_SYMBOL_GPL(ocelot_lag_fdb_del);
+
 /* Configure the maximum SDU (L2 payload) on RX to the value specified in @sdu.
  * The length of VLAN tags is accounted for automatically via DEV_MAC_TAGS_CFG.
  * In the special case that it's the NPI port that we're configuring, the
@@ -2769,6 +2894,7 @@ int ocelot_init(struct ocelot *ocelot)
 	INIT_LIST_HEAD(&ocelot->multicast);
 	INIT_LIST_HEAD(&ocelot->pgids);
 	INIT_LIST_HEAD(&ocelot->vlans);
+	INIT_LIST_HEAD(&ocelot->lag_fdbs);
 	ocelot_detect_features(ocelot);
 	ocelot_mact_init(ocelot);
 	ocelot_vlan_init(ocelot);
diff --git a/include/soc/mscc/ocelot.h b/include/soc/mscc/ocelot.h
index 78f56502bc09..dd4fc34d2992 100644
--- a/include/soc/mscc/ocelot.h
+++ b/include/soc/mscc/ocelot.h
@@ -635,6 +635,13 @@ enum macaccess_entry_type {
 #define OCELOT_QUIRK_PCS_PERFORMS_RATE_ADAPTATION	BIT(0)
 #define OCELOT_QUIRK_QSGMII_PORTS_MUST_BE_UP		BIT(1)
 
+struct ocelot_lag_fdb {
+	unsigned char addr[ETH_ALEN];
+	u16 vid;
+	struct net_device *bond;
+	struct list_head list;
+};
+
 struct ocelot_port {
 	struct ocelot			*ocelot;
 
@@ -690,6 +697,7 @@ struct ocelot {
 
 	struct list_head		vlans;
 	struct list_head		traps;
+	struct list_head		lag_fdbs;
 
 	/* Switches like VSC9959 have flooding per traffic class */
 	int				num_flooding_pgids;
@@ -866,6 +874,10 @@ int ocelot_fdb_add(struct ocelot *ocelot, int port,
 		   const unsigned char *addr, u16 vid);
 int ocelot_fdb_del(struct ocelot *ocelot, int port,
 		   const unsigned char *addr, u16 vid);
+int ocelot_lag_fdb_add(struct ocelot *ocelot, struct net_device *bond,
+		       const unsigned char *addr, u16 vid);
+int ocelot_lag_fdb_del(struct ocelot *ocelot, struct net_device *bond,
+		       const unsigned char *addr, u16 vid);
 int ocelot_vlan_prepare(struct ocelot *ocelot, int port, u16 vid, bool pvid,
 			bool untagged, struct netlink_ext_ack *extack);
 int ocelot_vlan_add(struct ocelot *ocelot, int port, u16 vid, bool pvid,
-- 
2.25.1