Re: [PATCH net-next 1/6] net: dsa: Fix SWITCHDEV_ATTR_ID_PORT_PARENT_ID

[Vivien Didelot]

Hi Andrew,

Andrew Lunn <and...@lunn.ch> writes:

> @@ -354,13 +354,16 @@ static int dsa_slave_port_attr_get(struct net_device 
> *dev,
>                                  struct switchdev_attr *attr)
>  {
>       struct dsa_slave_priv *p = netdev_priv(dev);
> -     struct dsa_switch *ds = p->dp->ds;
>  
>       switch (attr->id) {
> -     case SWITCHDEV_ATTR_ID_PORT_PARENT_ID:
> -             attr->u.ppid.id_len = sizeof(ds->index);
> -             memcpy(&attr->u.ppid.id, &ds->index, attr->u.ppid.id_len);
> +     case SWITCHDEV_ATTR_ID_PORT_PARENT_ID: {
> +             struct dsa_switch *ds = p->dp->ds;
> +             struct dsa_switch_tree *dst = ds->dst;
> +
> +             attr->u.ppid.id_len = sizeof(dst->tree);
> +             memcpy(&attr->u.ppid.id, &dst->tree, sizeof(dst->tree));
>               break;
> +     }
>       case SWITCHDEV_ATTR_ID_PORT_BRIDGE_FLAGS_SUPPORT:
>               attr->u.brport_flags_support = 0;
>               break;

I am pretty sure the kernel folks won't like blocks within case
statments. Simply declare dst = p->dp->ds->dst at the beginning.

Also keeping attr->u.ppid.id_len as memcpy's third argument like other
switchdev users do would be prefered.

Otherwise using the tree index is indeed the way to go:

Reviewed-by: Vivien Didelot <vivien.dide...@savoirfairelinux.com>


Thanks,

        Vivien