Re: [PATCH v2 net-next 02/10] net: bridge: mst: Allow changing a VLAN's MSTI

[Vladimir Oltean]

On Tue, Mar 01, 2022 at 11:03:13AM +0100, Tobias Waldekranz wrote:
> Allow a VLAN to move out of the CST (MSTI 0), to an independent tree.
> 
> The user manages the VID to MSTI mappings via a global VLAN
> setting. The proposed iproute2 interface would be:
> 
>     bridge vlan global set dev br0 vid <VID> msti <MSTI>
> 
> Changing the state in non-zero MSTIs is still not supported, but will
> be addressed in upcoming changes.
> 
> Signed-off-by: Tobias Waldekranz <tobias@xxxxxxxxxxxxxx>
> ---

> +static void br_mst_vlan_sync_state(struct net_bridge_vlan *pv, u16 msti)
> +{
> +	struct net_bridge_vlan_group *vg = nbp_vlan_group(pv->port);
> +	struct net_bridge_vlan *v;
> +
> +	list_for_each_entry(v, &vg->vlan_list, vlist) {
> +		/* If this port already has a defined state in this
> +		 * MSTI (through some other VLAN membership), inherit
> +		 * it.
> +		 */
> +		if (v != pv && v->brvlan->msti == msti) {
> +			br_mst_vlan_set_state(pv->port, pv, v->state);
> +			return;
> +		}
> +	}
> +
> +	/* Otherwise, start out in a new MSTI with all ports disabled. */
> +	return br_mst_vlan_set_state(pv->port, pv, BR_STATE_DISABLED);
> +}
> +
> +int br_mst_vlan_set_msti(struct net_bridge_vlan *mv, u16 msti)
> +{
> +	struct net_bridge_vlan_group *vg;
> +	struct net_bridge_vlan *pv;
> +	struct net_bridge_port *p;

No attempt to detect non-changes to the MSTI, and exit early? In a later
patch you will also notify switchdev uselessly because of this.

> +
> +	mv->msti = msti;
> +
> +	list_for_each_entry(p, &mv->br->port_list, list) {
> +		vg = nbp_vlan_group(p);
> +
> +		pv = br_vlan_find(vg, mv->vid);
> +		if (pv)
> +			br_mst_vlan_sync_state(pv, msti);
> +	}
> +
> +	return 0;
> +}