Merge branch 'mlxsw-resil-nexthop-groups-prep'

Ido Schimmel says:

====================
mlxsw: Preparations for resilient nexthop groups

This patchset contains preparations for resilient nexthop groups support in
mlxsw. A follow-up patchset will add support and selftests. Most of the
patches are trivial and small to make review easier.

Patchset overview:

Patch #1 removes RTNL assertion in nexthop notifier block since it is
not needed. The assertion will trigger when mlxsw starts processing
notifications related to resilient groups as not all are emitted with
RTNL held.

Patches #2-#9 gradually add support for nexthops with trap action. Up
until now mlxsw did not program nexthops whose neighbour entry was not
resolved. This will not work with resilient groups as their size is
fixed and the nexthop mapped to each bucket is determined by the nexthop
code. Therefore, nexthops whose neighbour entry is not resolved will be
programmed to trap packets to the CPU in order to trigger neighbour
resolution.

Patch #10 is a non-functional change to allow for code reuse between
regular nexthop groups and resilient ones.

Patch #11 avoids unnecessary neighbour updates in hardware. See the
commit message for a detailed explanation.

Patches #12-#14 add support for additional nexthop group sizes that are
supported by Spectrum-{2,3} ASICs.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/mellanox/mlxsw/spectrum.c

@@ -2934,7 +2934,6 @@ static int mlxsw_sp1_init(struct mlxsw_core *mlxsw_core,
 	mlxsw_sp->acl_tcam_ops = &mlxsw_sp1_acl_tcam_ops;
 	mlxsw_sp->nve_ops_arr = mlxsw_sp1_nve_ops_arr;
 	mlxsw_sp->mac_mask = mlxsw_sp1_mac_mask;
-	mlxsw_sp->rif_ops_arr = mlxsw_sp1_rif_ops_arr;
 	mlxsw_sp->sb_vals = &mlxsw_sp1_sb_vals;
 	mlxsw_sp->sb_ops = &mlxsw_sp1_sb_ops;
 	mlxsw_sp->port_type_speed_ops = &mlxsw_sp1_port_type_speed_ops;
@@ -2943,6 +2942,7 @@ static int mlxsw_sp1_init(struct mlxsw_core *mlxsw_core,
 	mlxsw_sp->policer_core_ops = &mlxsw_sp1_policer_core_ops;
 	mlxsw_sp->trap_ops = &mlxsw_sp1_trap_ops;
 	mlxsw_sp->mall_ops = &mlxsw_sp1_mall_ops;
+	mlxsw_sp->router_ops = &mlxsw_sp1_router_ops;
 	mlxsw_sp->listeners = mlxsw_sp1_listener;
 	mlxsw_sp->listeners_count = ARRAY_SIZE(mlxsw_sp1_listener);
 	mlxsw_sp->lowest_shaper_bs = MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP1;
@@ -2965,7 +2965,6 @@ static int mlxsw_sp2_init(struct mlxsw_core *mlxsw_core,
 	mlxsw_sp->acl_tcam_ops = &mlxsw_sp2_acl_tcam_ops;
 	mlxsw_sp->nve_ops_arr = mlxsw_sp2_nve_ops_arr;
 	mlxsw_sp->mac_mask = mlxsw_sp2_mac_mask;
-	mlxsw_sp->rif_ops_arr = mlxsw_sp2_rif_ops_arr;
 	mlxsw_sp->sb_vals = &mlxsw_sp2_sb_vals;
 	mlxsw_sp->sb_ops = &mlxsw_sp2_sb_ops;
 	mlxsw_sp->port_type_speed_ops = &mlxsw_sp2_port_type_speed_ops;
@@ -2974,6 +2973,7 @@ static int mlxsw_sp2_init(struct mlxsw_core *mlxsw_core,
 	mlxsw_sp->policer_core_ops = &mlxsw_sp2_policer_core_ops;
 	mlxsw_sp->trap_ops = &mlxsw_sp2_trap_ops;
 	mlxsw_sp->mall_ops = &mlxsw_sp2_mall_ops;
+	mlxsw_sp->router_ops = &mlxsw_sp2_router_ops;
 	mlxsw_sp->lowest_shaper_bs = MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP2;
 
 	return mlxsw_sp_init(mlxsw_core, mlxsw_bus_info, extack);
@@ -2994,7 +2994,6 @@ static int mlxsw_sp3_init(struct mlxsw_core *mlxsw_core,
 	mlxsw_sp->acl_tcam_ops = &mlxsw_sp2_acl_tcam_ops;
 	mlxsw_sp->nve_ops_arr = mlxsw_sp2_nve_ops_arr;
 	mlxsw_sp->mac_mask = mlxsw_sp2_mac_mask;
-	mlxsw_sp->rif_ops_arr = mlxsw_sp2_rif_ops_arr;
 	mlxsw_sp->sb_vals = &mlxsw_sp2_sb_vals;
 	mlxsw_sp->sb_ops = &mlxsw_sp3_sb_ops;
 	mlxsw_sp->port_type_speed_ops = &mlxsw_sp2_port_type_speed_ops;
@@ -3003,6 +3002,7 @@ static int mlxsw_sp3_init(struct mlxsw_core *mlxsw_core,
 	mlxsw_sp->policer_core_ops = &mlxsw_sp2_policer_core_ops;
 	mlxsw_sp->trap_ops = &mlxsw_sp2_trap_ops;
 	mlxsw_sp->mall_ops = &mlxsw_sp2_mall_ops;
+	mlxsw_sp->router_ops = &mlxsw_sp2_router_ops;
 	mlxsw_sp->lowest_shaper_bs = MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP3;
 
 	return mlxsw_sp_init(mlxsw_core, mlxsw_bus_info, extack);

drivers/net/ethernet/mellanox/mlxsw/spectrum.h

@@ -87,10 +87,10 @@ enum mlxsw_sp_rif_type {
 	MLXSW_SP_RIF_TYPE_MAX,
 };
 
-struct mlxsw_sp_rif_ops;
+struct mlxsw_sp_router_ops;
 
-extern const struct mlxsw_sp_rif_ops *mlxsw_sp1_rif_ops_arr[];
-extern const struct mlxsw_sp_rif_ops *mlxsw_sp2_rif_ops_arr[];
+extern const struct mlxsw_sp_router_ops mlxsw_sp1_router_ops;
+extern const struct mlxsw_sp_router_ops mlxsw_sp2_router_ops;
 
 struct mlxsw_sp_switchdev_ops;
 
@@ -180,7 +180,6 @@ struct mlxsw_sp {
 	const struct mlxsw_sp_acl_rulei_ops *acl_rulei_ops;
 	const struct mlxsw_sp_acl_tcam_ops *acl_tcam_ops;
 	const struct mlxsw_sp_nve_ops **nve_ops_arr;
-	const struct mlxsw_sp_rif_ops **rif_ops_arr;
 	const struct mlxsw_sp_sb_vals *sb_vals;
 	const struct mlxsw_sp_sb_ops *sb_ops;
 	const struct mlxsw_sp_port_type_speed_ops *port_type_speed_ops;
@@ -189,6 +188,7 @@ struct mlxsw_sp {
 	const struct mlxsw_sp_policer_core_ops *policer_core_ops;
 	const struct mlxsw_sp_trap_ops *trap_ops;
 	const struct mlxsw_sp_mall_ops *mall_ops;
+	const struct mlxsw_sp_router_ops *router_ops;
 	const struct mlxsw_listener *listeners;
 	size_t listeners_count;
 	u32 lowest_shaper_bs;

drivers/net/ethernet/mellanox/mlxsw/spectrum_dpipe.c

@@ -912,9 +912,8 @@ static u64 mlxsw_sp_dpipe_table_adj_size(struct mlxsw_sp *mlxsw_sp)
 	u64 size = 0;
 
 	mlxsw_sp_nexthop_for_each(nh, mlxsw_sp->router)
-		if (mlxsw_sp_nexthop_offload(nh) &&
-		    !mlxsw_sp_nexthop_group_has_ipip(nh) &&
-		    !mlxsw_sp_nexthop_is_discard(nh))
+		if (mlxsw_sp_nexthop_is_forward(nh) &&
+		    !mlxsw_sp_nexthop_group_has_ipip(nh))
 			size++;
 	return size;
 }
@@ -1105,9 +1104,8 @@ mlxsw_sp_dpipe_table_adj_entries_get(struct mlxsw_sp *mlxsw_sp,
 	nh_skip = nh_count;
 	nh_count = 0;
 	mlxsw_sp_nexthop_for_each(nh, mlxsw_sp->router) {
-		if (!mlxsw_sp_nexthop_offload(nh) ||
-		    mlxsw_sp_nexthop_group_has_ipip(nh) ||
-		    mlxsw_sp_nexthop_is_discard(nh))
+		if (!mlxsw_sp_nexthop_is_forward(nh) ||
+		    mlxsw_sp_nexthop_group_has_ipip(nh))
 			continue;
 
 		if (nh_count < nh_skip)
@@ -1187,9 +1185,8 @@ static int mlxsw_sp_dpipe_table_adj_counters_update(void *priv, bool enable)
 	u32 adj_size = 0;
 
 	mlxsw_sp_nexthop_for_each(nh, mlxsw_sp->router) {
-		if (!mlxsw_sp_nexthop_offload(nh) ||
-		    mlxsw_sp_nexthop_group_has_ipip(nh) ||
-		    mlxsw_sp_nexthop_is_discard(nh))
+		if (!mlxsw_sp_nexthop_is_forward(nh) ||
+		    mlxsw_sp_nexthop_group_has_ipip(nh))
 			continue;
 
 		mlxsw_sp_nexthop_indexes(nh, &adj_index, &adj_size,
@@ -1198,8 +1195,8 @@ static int mlxsw_sp_dpipe_table_adj_counters_update(void *priv, bool enable)
 			mlxsw_sp_nexthop_counter_alloc(mlxsw_sp, nh);
 		else
 			mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
-		mlxsw_sp_nexthop_update(mlxsw_sp,
-					adj_index + adj_hash_index, nh);
+		mlxsw_sp_nexthop_eth_update(mlxsw_sp,
+					    adj_index + adj_hash_index, nh);
 	}
 	return 0;
 }

drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c

@@ -113,6 +113,10 @@ struct mlxsw_sp_rif_ops {
 	void (*fdb_del)(struct mlxsw_sp_rif *rif, const char *mac);
 };
 
+struct mlxsw_sp_router_ops {
+	int (*init)(struct mlxsw_sp *mlxsw_sp);
+};
+
 static struct mlxsw_sp_rif *
 mlxsw_sp_rif_find_by_dev(const struct mlxsw_sp *mlxsw_sp,
 			 const struct net_device *dev);
@@ -2662,6 +2666,10 @@ static void mlxsw_sp_router_neigh_event_work(struct work_struct *work)
 			goto out;
 	}
 
+	if (neigh_entry->connected && entry_connected &&
+	    !memcmp(neigh_entry->ha, ha, ETH_ALEN))
+		goto out;
+
 	memcpy(neigh_entry->ha, ha, ETH_ALEN);
 	mlxsw_sp_neigh_entry_update(mlxsw_sp, neigh_entry, entry_connected);
 	mlxsw_sp_nexthop_neigh_update(mlxsw_sp, neigh_entry, !entry_connected,
@@ -2842,6 +2850,15 @@ enum mlxsw_sp_nexthop_type {
 	MLXSW_SP_NEXTHOP_TYPE_IPIP,
 };
 
+enum mlxsw_sp_nexthop_action {
+	/* Nexthop forwards packets to an egress RIF */
+	MLXSW_SP_NEXTHOP_ACTION_FORWARD,
+	/* Nexthop discards packets */
+	MLXSW_SP_NEXTHOP_ACTION_DISCARD,
+	/* Nexthop traps packets */
+	MLXSW_SP_NEXTHOP_ACTION_TRAP,
+};
+
 struct mlxsw_sp_nexthop_key {
 	struct fib_nh *fib_nh;
 };
@@ -2862,16 +2879,16 @@ struct mlxsw_sp_nexthop {
 	int norm_nh_weight;
 	int num_adj_entries;
 	struct mlxsw_sp_rif *rif;
-	u8 should_offload:1, /* set indicates this neigh is connected and
-			      * should be put to KVD linear area of this group.
+	u8 should_offload:1, /* set indicates this nexthop should be written
+			      * to the adjacency table.
 			      */
-	   offloaded:1, /* set in case the neigh is actually put into
-			 * KVD linear area of this group.
+	   offloaded:1, /* set indicates this nexthop was written to the
+			 * adjacency table.
 			 */
-	   update:1, /* set indicates that MAC of this neigh should be
-		      * updated in HW
+	   update:1; /* set indicates this nexthop should be updated in the
+		      * adjacency table (f.e., its MAC changed).
 		      */
-	   discard:1; /* nexthop is programmed to discard packets */
+	enum mlxsw_sp_nexthop_action action;
 	enum mlxsw_sp_nexthop_type type;
 	union {
 		struct mlxsw_sp_neigh_entry *neigh_entry;
@@ -2979,14 +2996,15 @@ struct mlxsw_sp_nexthop *mlxsw_sp_nexthop_next(struct mlxsw_sp_router *router,
 	return list_next_entry(nh, router_list_node);
 }
 
-bool mlxsw_sp_nexthop_offload(struct mlxsw_sp_nexthop *nh)
+bool mlxsw_sp_nexthop_is_forward(const struct mlxsw_sp_nexthop *nh)
 {
-	return nh->offloaded;
+	return nh->offloaded && nh->action == MLXSW_SP_NEXTHOP_ACTION_FORWARD;
 }
 
 unsigned char *mlxsw_sp_nexthop_ha(struct mlxsw_sp_nexthop *nh)
 {
-	if (!nh->offloaded)
+	if (nh->type != MLXSW_SP_NEXTHOP_TYPE_ETH ||
+	    !mlxsw_sp_nexthop_is_forward(nh))
 		return NULL;
 	return nh->neigh_entry->ha;
 }
@@ -3036,11 +3054,6 @@ bool mlxsw_sp_nexthop_group_has_ipip(struct mlxsw_sp_nexthop *nh)
 	return false;
 }
 
-bool mlxsw_sp_nexthop_is_discard(const struct mlxsw_sp_nexthop *nh)
-{
-	return nh->discard;
-}
-
 static const struct rhashtable_params mlxsw_sp_nexthop_group_vr_ht_params = {
 	.key_offset = offsetof(struct mlxsw_sp_nexthop_group_vr_entry, key),
 	.head_offset = offsetof(struct mlxsw_sp_nexthop_group_vr_entry, ht_node),
@@ -3403,20 +3416,36 @@ static int mlxsw_sp_adj_index_mass_update(struct mlxsw_sp *mlxsw_sp,
 	return err;
 }
 
-static int __mlxsw_sp_nexthop_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
-				     struct mlxsw_sp_nexthop *nh)
+static int __mlxsw_sp_nexthop_eth_update(struct mlxsw_sp *mlxsw_sp,
+					 u32 adj_index,
+					 struct mlxsw_sp_nexthop *nh)
 {
 	struct mlxsw_sp_neigh_entry *neigh_entry = nh->neigh_entry;
 	char ratr_pl[MLXSW_REG_RATR_LEN];
+	u16 rif_index;
 
+	rif_index = nh->rif ? nh->rif->rif_index :
+			      mlxsw_sp->router->lb_rif_index;
 	mlxsw_reg_ratr_pack(ratr_pl, MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY,
 			    true, MLXSW_REG_RATR_TYPE_ETHERNET,
-			    adj_index, nh->rif->rif_index);
-	if (nh->discard)
+			    adj_index, rif_index);
+	switch (nh->action) {
+	case MLXSW_SP_NEXTHOP_ACTION_FORWARD:
+		mlxsw_reg_ratr_eth_entry_pack(ratr_pl, neigh_entry->ha);
+		break;
+	case MLXSW_SP_NEXTHOP_ACTION_DISCARD:
 		mlxsw_reg_ratr_trap_action_set(ratr_pl,
 					       MLXSW_REG_RATR_TRAP_ACTION_DISCARD_ERRORS);
-	else
-		mlxsw_reg_ratr_eth_entry_pack(ratr_pl, neigh_entry->ha);
+		break;
+	case MLXSW_SP_NEXTHOP_ACTION_TRAP:
+		mlxsw_reg_ratr_trap_action_set(ratr_pl,
+					       MLXSW_REG_RATR_TRAP_ACTION_TRAP);
+		mlxsw_reg_ratr_trap_id_set(ratr_pl, MLXSW_TRAP_ID_RTR_EGRESS0);
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
 	if (nh->counter_valid)
 		mlxsw_reg_ratr_counter_pack(ratr_pl, nh->counter_index, true);
 	else
@@ -3425,15 +3454,16 @@ static int __mlxsw_sp_nexthop_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ratr), ratr_pl);
 }
 
-int mlxsw_sp_nexthop_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
-			    struct mlxsw_sp_nexthop *nh)
+int mlxsw_sp_nexthop_eth_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				struct mlxsw_sp_nexthop *nh)
 {
 	int i;
 
 	for (i = 0; i < nh->num_adj_entries; i++) {
 		int err;
 
-		err = __mlxsw_sp_nexthop_update(mlxsw_sp, adj_index + i, nh);
+		err = __mlxsw_sp_nexthop_eth_update(mlxsw_sp, adj_index + i,
+						    nh);
 		if (err)
 			return err;
 	}
@@ -3469,6 +3499,20 @@ static int mlxsw_sp_nexthop_ipip_update(struct mlxsw_sp *mlxsw_sp,
 	return 0;
 }
 
+static int mlxsw_sp_nexthop_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				   struct mlxsw_sp_nexthop *nh)
+{
+	/* When action is discard or trap, the nexthop must be
+	 * programmed as an Ethernet nexthop.
+	 */
+	if (nh->type == MLXSW_SP_NEXTHOP_TYPE_ETH ||
+	    nh->action == MLXSW_SP_NEXTHOP_ACTION_DISCARD ||
+	    nh->action == MLXSW_SP_NEXTHOP_ACTION_TRAP)
+		return mlxsw_sp_nexthop_eth_update(mlxsw_sp, adj_index, nh);
+	else
+		return mlxsw_sp_nexthop_ipip_update(mlxsw_sp, adj_index, nh);
+}
+
 static int
 mlxsw_sp_nexthop_group_update(struct mlxsw_sp *mlxsw_sp,
 			      struct mlxsw_sp_nexthop_group_info *nhgi,
@@ -3489,16 +3533,7 @@ mlxsw_sp_nexthop_group_update(struct mlxsw_sp *mlxsw_sp,
 		if (nh->update || reallocate) {
 			int err = 0;
 
-			switch (nh->type) {
-			case MLXSW_SP_NEXTHOP_TYPE_ETH:
-				err = mlxsw_sp_nexthop_update
-					    (mlxsw_sp, adj_index, nh);
-				break;
-			case MLXSW_SP_NEXTHOP_TYPE_IPIP:
-				err = mlxsw_sp_nexthop_ipip_update
-					    (mlxsw_sp, adj_index, nh);
-				break;
-			}
+			err = mlxsw_sp_nexthop_update(mlxsw_sp, adj_index, nh);
 			if (err)
 				return err;
 			nh->update = 0;
@@ -3524,34 +3559,69 @@ mlxsw_sp_nexthop_fib_entries_update(struct mlxsw_sp *mlxsw_sp,
 	return 0;
 }
 
-static void mlxsw_sp_adj_grp_size_round_up(u16 *p_adj_grp_size)
+struct mlxsw_sp_adj_grp_size_range {
+	u16 start; /* Inclusive */
+	u16 end; /* Inclusive */
+};
+
+/* Ordered by range start value */
+static const struct mlxsw_sp_adj_grp_size_range
+mlxsw_sp1_adj_grp_size_ranges[] = {
+	{ .start = 1, .end = 64 },
+	{ .start = 512, .end = 512 },
+	{ .start = 1024, .end = 1024 },
+	{ .start = 2048, .end = 2048 },
+	{ .start = 4096, .end = 4096 },
+};
+
+/* Ordered by range start value */
+static const struct mlxsw_sp_adj_grp_size_range
+mlxsw_sp2_adj_grp_size_ranges[] = {
+	{ .start = 1, .end = 128 },
+	{ .start = 256, .end = 256 },
+	{ .start = 512, .end = 512 },
+	{ .start = 1024, .end = 1024 },
+	{ .start = 2048, .end = 2048 },
+	{ .start = 4096, .end = 4096 },
+};
+
+static void mlxsw_sp_adj_grp_size_round_up(const struct mlxsw_sp *mlxsw_sp,
+					   u16 *p_adj_grp_size)
 {
-	/* Valid sizes for an adjacency group are:
-	 * 1-64, 512, 1024, 2048 and 4096.
-	 */
-	if (*p_adj_grp_size <= 64)
-		return;
-	else if (*p_adj_grp_size <= 512)
-		*p_adj_grp_size = 512;
-	else if (*p_adj_grp_size <= 1024)
-		*p_adj_grp_size = 1024;
-	else if (*p_adj_grp_size <= 2048)
-		*p_adj_grp_size = 2048;
-	else
-		*p_adj_grp_size = 4096;
+	int i;
+
+	for (i = 0; i < mlxsw_sp->router->adj_grp_size_ranges_count; i++) {
+		const struct mlxsw_sp_adj_grp_size_range *size_range;
+
+		size_range = &mlxsw_sp->router->adj_grp_size_ranges[i];
+
+		if (*p_adj_grp_size >= size_range->start &&
+		    *p_adj_grp_size <= size_range->end)
+			return;
+
+		if (*p_adj_grp_size <= size_range->end) {
+			*p_adj_grp_size = size_range->end;
+			return;
+		}
+	}
 }
 
-static void mlxsw_sp_adj_grp_size_round_down(u16 *p_adj_grp_size,
+static void mlxsw_sp_adj_grp_size_round_down(const struct mlxsw_sp *mlxsw_sp,
+					     u16 *p_adj_grp_size,
 					     unsigned int alloc_size)
 {
-	if (alloc_size >= 4096)
-		*p_adj_grp_size = 4096;
-	else if (alloc_size >= 2048)
-		*p_adj_grp_size = 2048;
-	else if (alloc_size >= 1024)
-		*p_adj_grp_size = 1024;
-	else if (alloc_size >= 512)
-		*p_adj_grp_size = 512;
+	int i;
+
+	for (i = mlxsw_sp->router->adj_grp_size_ranges_count - 1; i >= 0; i--) {
+		const struct mlxsw_sp_adj_grp_size_range *size_range;
+
+		size_range = &mlxsw_sp->router->adj_grp_size_ranges[i];
+
+		if (alloc_size >= size_range->end) {
+			*p_adj_grp_size = size_range->end;
+			return;
+		}
+	}
 }
 
 static int mlxsw_sp_fix_adj_grp_size(struct mlxsw_sp *mlxsw_sp,
@@ -3563,7 +3633,7 @@ static int mlxsw_sp_fix_adj_grp_size(struct mlxsw_sp *mlxsw_sp,
 	/* Round up the requested group size to the next size supported
 	 * by the device and make sure the request can be satisfied.
 	 */
-	mlxsw_sp_adj_grp_size_round_up(p_adj_grp_size);
+	mlxsw_sp_adj_grp_size_round_up(mlxsw_sp, p_adj_grp_size);
 	err = mlxsw_sp_kvdl_alloc_count_query(mlxsw_sp,
 					      MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
 					      *p_adj_grp_size, &alloc_size);
@@ -3573,7 +3643,7 @@ static int mlxsw_sp_fix_adj_grp_size(struct mlxsw_sp *mlxsw_sp,
 	 * entries than requested. Try to use as much of them as
 	 * possible.
 	 */
-	mlxsw_sp_adj_grp_size_round_down(p_adj_grp_size, alloc_size);
+	mlxsw_sp_adj_grp_size_round_down(mlxsw_sp, p_adj_grp_size, alloc_size);
 
 	return 0;
 }
@@ -3832,10 +3902,12 @@ mlxsw_sp_nexthop_group_refresh(struct mlxsw_sp *mlxsw_sp,
 static void __mlxsw_sp_nexthop_neigh_update(struct mlxsw_sp_nexthop *nh,
 					    bool removing)
 {
-	if (!removing)
+	if (!removing) {
+		nh->action = MLXSW_SP_NEXTHOP_ACTION_FORWARD;
 		nh->should_offload = 1;
-	else
+	} else {
 		nh->should_offload = 0;
+	}
 	nh->update = 1;
 }
 
@@ -4267,6 +4339,29 @@ mlxsw_sp_nexthop_obj_single_validate(struct mlxsw_sp *mlxsw_sp,
 	return err;
 }
 
+static int
+mlxsw_sp_nexthop_obj_group_entry_validate(struct mlxsw_sp *mlxsw_sp,
+					  const struct nh_notifier_single_info *nh,
+					  struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlxsw_sp_nexthop_obj_single_validate(mlxsw_sp, nh, extack);
+	if (err)
+		return err;
+
+	/* Device only nexthops with an IPIP device are programmed as
+	 * encapsulating adjacency entries.
+	 */
+	if (!nh->gw_family && !nh->is_reject &&
+	    !mlxsw_sp_netdev_ipip_type(mlxsw_sp, nh->dev, NULL)) {
+		NL_SET_ERR_MSG_MOD(extack, "Nexthop group entry does not have a gateway");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 static int
 mlxsw_sp_nexthop_obj_group_validate(struct mlxsw_sp *mlxsw_sp,
 				    const struct nh_notifier_grp_info *nh_grp,
@@ -4284,19 +4379,10 @@ mlxsw_sp_nexthop_obj_group_validate(struct mlxsw_sp *mlxsw_sp,
 		int err;
 
 		nh = &nh_grp->nh_entries[i].nh;
-		err = mlxsw_sp_nexthop_obj_single_validate(mlxsw_sp, nh,
-							   extack);
+		err = mlxsw_sp_nexthop_obj_group_entry_validate(mlxsw_sp, nh,
+								extack);
 		if (err)
 			return err;
-
-		/* Device only nexthops with an IPIP device are programmed as
-		 * encapsulating adjacency entries.
-		 */
-		if (!nh->gw_family && !nh->is_reject &&
-		    !mlxsw_sp_netdev_ipip_type(mlxsw_sp, nh->dev, NULL)) {
-			NL_SET_ERR_MSG_MOD(extack, "Nexthop group entry does not have a gateway");
-			return -EINVAL;
-		}
 	}
 
 	return 0;
@@ -4346,7 +4432,7 @@ static void mlxsw_sp_nexthop_obj_blackhole_init(struct mlxsw_sp *mlxsw_sp,
 {
 	u16 lb_rif_index = mlxsw_sp->router->lb_rif_index;
 
-	nh->discard = 1;
+	nh->action = MLXSW_SP_NEXTHOP_ACTION_DISCARD;
 	nh->should_offload = 1;
 	/* While nexthops that discard packets do not forward packets
 	 * via an egress RIF, they still need to be programmed using a
@@ -4409,7 +4495,7 @@ mlxsw_sp_nexthop_obj_init(struct mlxsw_sp *mlxsw_sp,
 static void mlxsw_sp_nexthop_obj_fini(struct mlxsw_sp *mlxsw_sp,
 				      struct mlxsw_sp_nexthop *nh)
 {
-	if (nh->discard)
+	if (nh->action == MLXSW_SP_NEXTHOP_ACTION_DISCARD)
 		mlxsw_sp_nexthop_obj_blackhole_fini(mlxsw_sp, nh);
 	mlxsw_sp_nexthop_type_fini(mlxsw_sp, nh);
 	list_del(&nh->router_list_node);
@@ -4699,8 +4785,6 @@ static int mlxsw_sp_nexthop_obj_event(struct notifier_block *nb,
 
 	mutex_lock(&router->lock);
 
-	ASSERT_RTNL();
-
 	switch (event) {
 	case NEXTHOP_EVENT_REPLACE:
 		err = mlxsw_sp_nexthop_obj_new(router->mlxsw_sp, info);
@@ -7667,7 +7751,7 @@ mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
 	int i, err;
 
 	type = mlxsw_sp_dev_rif_type(mlxsw_sp, params->dev);
-	ops = mlxsw_sp->rif_ops_arr[type];
+	ops = mlxsw_sp->router->rif_ops_arr[type];
 
 	vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id ? : RT_TABLE_MAIN, extack);
 	if (IS_ERR(vr))
@@ -8865,7 +8949,7 @@ static const struct mlxsw_sp_rif_ops mlxsw_sp1_rif_ipip_lb_ops = {
 	.deconfigure		= mlxsw_sp1_rif_ipip_lb_deconfigure,
 };
 
-const struct mlxsw_sp_rif_ops *mlxsw_sp1_rif_ops_arr[] = {
+static const struct mlxsw_sp_rif_ops *mlxsw_sp1_rif_ops_arr[] = {
 	[MLXSW_SP_RIF_TYPE_SUBPORT]	= &mlxsw_sp_rif_subport_ops,
 	[MLXSW_SP_RIF_TYPE_VLAN]	= &mlxsw_sp_rif_vlan_emu_ops,
 	[MLXSW_SP_RIF_TYPE_FID]		= &mlxsw_sp_rif_fid_ops,
@@ -9050,7 +9134,7 @@ static const struct mlxsw_sp_rif_ops mlxsw_sp2_rif_ipip_lb_ops = {
 	.deconfigure		= mlxsw_sp2_rif_ipip_lb_deconfigure,
 };
 
-const struct mlxsw_sp_rif_ops *mlxsw_sp2_rif_ops_arr[] = {
+static const struct mlxsw_sp_rif_ops *mlxsw_sp2_rif_ops_arr[] = {
 	[MLXSW_SP_RIF_TYPE_SUBPORT]	= &mlxsw_sp_rif_subport_ops,
 	[MLXSW_SP_RIF_TYPE_VLAN]	= &mlxsw_sp_rif_vlan_emu_ops,
 	[MLXSW_SP_RIF_TYPE_FID]		= &mlxsw_sp_rif_fid_ops,
@@ -9302,6 +9386,36 @@ static void mlxsw_sp_lb_rif_fini(struct mlxsw_sp *mlxsw_sp)
 	mlxsw_sp_router_ul_rif_put(mlxsw_sp, mlxsw_sp->router->lb_rif_index);
 }
 
+static int mlxsw_sp1_router_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t size_ranges_count = ARRAY_SIZE(mlxsw_sp1_adj_grp_size_ranges);
+
+	mlxsw_sp->router->rif_ops_arr = mlxsw_sp1_rif_ops_arr;
+	mlxsw_sp->router->adj_grp_size_ranges = mlxsw_sp1_adj_grp_size_ranges;
+	mlxsw_sp->router->adj_grp_size_ranges_count = size_ranges_count;
+
+	return 0;
+}
+
+const struct mlxsw_sp_router_ops mlxsw_sp1_router_ops = {
+	.init = mlxsw_sp1_router_init,
+};
+
+static int mlxsw_sp2_router_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t size_ranges_count = ARRAY_SIZE(mlxsw_sp2_adj_grp_size_ranges);
+
+	mlxsw_sp->router->rif_ops_arr = mlxsw_sp2_rif_ops_arr;
+	mlxsw_sp->router->adj_grp_size_ranges = mlxsw_sp2_adj_grp_size_ranges;
+	mlxsw_sp->router->adj_grp_size_ranges_count = size_ranges_count;
+
+	return 0;
+}
+
+const struct mlxsw_sp_router_ops mlxsw_sp2_router_ops = {
+	.init = mlxsw_sp2_router_init,
+};
+
 int mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp,
 			 struct netlink_ext_ack *extack)
 {
@@ -9315,6 +9429,10 @@ int mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp,
 	mlxsw_sp->router = router;
 	router->mlxsw_sp = mlxsw_sp;
 
+	err = mlxsw_sp->router_ops->init(mlxsw_sp);
+	if (err)
+		goto err_router_ops_init;
+
 	err = mlxsw_sp_router_xm_init(mlxsw_sp);
 	if (err)
 		goto err_xm_init;
@@ -9455,6 +9573,7 @@ int mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp,
 err_ll_op_ctx_init:
 	mlxsw_sp_router_xm_fini(mlxsw_sp);
 err_xm_init:
+err_router_ops_init:
 	mutex_destroy(&mlxsw_sp->router->lock);
 	kfree(mlxsw_sp->router);
 	return err;

drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h

@@ -78,6 +78,8 @@ struct mlxsw_sp_router {
 	struct mlxsw_sp_fib_entry_op_ctx *ll_op_ctx;
 	u16 lb_rif_index;
 	struct mlxsw_sp_router_xm *xm;
+	const struct mlxsw_sp_adj_grp_size_range *adj_grp_size_ranges;
+	size_t adj_grp_size_ranges_count;
 };
 
 struct mlxsw_sp_fib_entry_priv {
@@ -195,20 +197,19 @@ mlxsw_sp_ipip_demote_tunnel_by_saddr(struct mlxsw_sp *mlxsw_sp,
 				     const struct mlxsw_sp_ipip_entry *except);
 struct mlxsw_sp_nexthop *mlxsw_sp_nexthop_next(struct mlxsw_sp_router *router,
 					       struct mlxsw_sp_nexthop *nh);
-bool mlxsw_sp_nexthop_offload(struct mlxsw_sp_nexthop *nh);
+bool mlxsw_sp_nexthop_is_forward(const struct mlxsw_sp_nexthop *nh);
 unsigned char *mlxsw_sp_nexthop_ha(struct mlxsw_sp_nexthop *nh);
 int mlxsw_sp_nexthop_indexes(struct mlxsw_sp_nexthop *nh, u32 *p_adj_index,
 			     u32 *p_adj_size, u32 *p_adj_hash_index);
 struct mlxsw_sp_rif *mlxsw_sp_nexthop_rif(struct mlxsw_sp_nexthop *nh);
 bool mlxsw_sp_nexthop_group_has_ipip(struct mlxsw_sp_nexthop *nh);
-bool mlxsw_sp_nexthop_is_discard(const struct mlxsw_sp_nexthop *nh);
 #define mlxsw_sp_nexthop_for_each(nh, router)				\
 	for (nh = mlxsw_sp_nexthop_next(router, NULL); nh;		\
 	     nh = mlxsw_sp_nexthop_next(router, nh))
 int mlxsw_sp_nexthop_counter_get(struct mlxsw_sp *mlxsw_sp,
 				 struct mlxsw_sp_nexthop *nh, u64 *p_counter);
-int mlxsw_sp_nexthop_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
-			    struct mlxsw_sp_nexthop *nh);
+int mlxsw_sp_nexthop_eth_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				struct mlxsw_sp_nexthop *nh);
 void mlxsw_sp_nexthop_counter_alloc(struct mlxsw_sp *mlxsw_sp,
 				    struct mlxsw_sp_nexthop *nh);
 void mlxsw_sp_nexthop_counter_free(struct mlxsw_sp *mlxsw_sp,