On Fri, May 13, 2016 at 07:07:52PM +0800, Marek Lindner wrote:
On Tuesday, May 10, 2016 18:06:08 Antonio Quartulli wrote:
+static struct batadv_gw_node * +batadv_iv_gw_get_best_gw_node(struct batadv_priv *bat_priv)
+static bool batadv_iv_gw_is_eligible(struct batadv_priv *bat_priv,
struct batadv_orig_node *curr_gw_orig,struct batadv_orig_node *orig_node)+/* fails if orig_node has no router */ +static int batadv_iv_gw_write_buffer_text(struct batadv_priv *bat_priv,
struct seq_file *seq,const struct batadv_gw_node *gw_node)+static void batadv_iv_gw_print(struct batadv_priv *bat_priv,
struct seq_file *seq)How about some kernel doc ? :)
ok :)
+211,7 @@ void batadv_gw_election(struct batadv_priv *bat_priv) if (!batadv_atomic_dec_not_zero(&bat_priv->gw.reselect) && curr_gw) goto out;
- next_gw = batadv_gw_get_best_gw_node(bat_priv);
- next_gw = bat_priv->algo_ops->gw.get_best_gw_node(bat_priv);
Either make the gw.get_best_gw_node() ops mandatory or check for its existence before using it.
I think we should always check the existence of this function.
However I also wanted to propose to reject any change of the GW mode if we do not have these APIs implemented. But tihs can be done with a later patch.
@@ -1320,11 +1320,21 @@ struct batadv_algo_orig_ops {
- struct batadv_algo_gw_ops - mesh algorithm callbacks (GW specific)
- @store_sel_class: parse and stores a new GW selection class
- @show_sel_class: prints the current GW selection class
- @get_best_gw_node: select the best GW from the list of available nodes
- @is_eligible: check if a newly discovered GW is a potential candidate
for
- the election as best GW
- @print: print the gateway table (optional)
*/
Shall we mark all optional ops as optional or .. ?
Dunno...this was a just a copy/paste from other "print" APIs. Maybe we just remove the "optional" text from here for now.
Cheers,