Stephen Hemminger <step...@networkplumber.org> writes:
> On Fri, 27 Apr 2018 18:11:06 +0300
> Ido Schimmel <ido...@mellanox.com> wrote:
>
>> +int br_vlan_pvid_rtnl(const struct net_device *dev, u16 *p_pvid)
>> +{
>> + struct net_bridge_vlan_group *vg;
>> +
>> + ASSERT_RTNL();
>> + if (netif_is_bridge_master(dev))
>> + vg = br_vlan_group(netdev_priv(dev));
>> + else
>> + return -EINVAL;
>> +
>> + *p_pvid = br_get_pvid(vg);
>> + return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(br_vlan_pvid_rtnl);
>
> Unless there is a RCU version, no need for _rtnl suffix.
All right, I wanted to be explicit, but if you prefer to drop the _rtnl,
no problem.
> Minor style issue, why not make error then go on and return -1 on error.
Basically to make it more straightforward to extend when people want to
add support for the other device kind (bridge / port).
>
>
> int br_vlan_pvid(const struct net_device *dev)
> {
> const struct net_bridge_vlan_group *vg;
>
> ASSERT_RTNL();
>
> if (!netif_is_bridge_master(dev))
> return -1;
>
> vg = br_vlan_group(netdev_priv(dev));
> return br_get_pvid(vg);
> }
