On Mon, Mar 14, 2022 at 10:52:20AM +0100, Tobias Waldekranz wrote:

+int br_mst_fill_info(struct sk_buff *skb, struct net_bridge_vlan_group *vg)
+{
+	struct net_bridge_vlan *v;
+	struct nlattr *nest;
+	unsigned long *seen;
+	int err = 0;
+
+	seen = bitmap_zalloc(VLAN_N_VID, 0);

I see there is precedent in the bridge driver for using dynamic
allocation as opposed to on-stack declaration using DECLARE_BITMAP().
I imagine this isn't just to be "heapsters", but why?

I don't have a very good sense of how much on-stack memory is too much
(a lot probably depends on the expected depth of the call stack too, and here it
doesn't appear to be too deep), but I see that mlxsw_sp_bridge_vxlan_vlan_is_valid()
has a DECLARE_BITMAP(vlans, VLAN_N_VID) too.

The comment applies for callers of br_mst_get_info() too.

+	if (!seen)
+		return -ENOMEM;
+
+	list_for_each_entry(v, &vg->vlan_list, vlist) {
+		if (test_bit(v->brvlan->msti, seen))
+			continue;
+
+		nest = nla_nest_start_noflag(skb, IFLA_BRIDGE_MST_ENTRY);
+		if (!nest ||
+		    nla_put_u16(skb, IFLA_BRIDGE_MST_ENTRY_MSTI, v->brvlan->msti) ||
+		    nla_put_u8(skb, IFLA_BRIDGE_MST_ENTRY_STATE, v->state)) {
+			err = -EMSGSIZE;
+			break;
+		}
+		nla_nest_end(skb, nest);
+
+		set_bit(v->brvlan->msti, seen);
+	}
+
+	kfree(seen);
+	return err;
+}