Re: [B.A.T.M.A.N.] batman-adv gateway handling

[Marek Lindner <lindner_marek@yahoo.de>]

On Saturday 30 October 2010 14:14:02 Sven Eckelmann wrote:
> > -       struct gw_node *gw_node;
> > -       struct hlist_node *node, *node_tmp;
> > +                                 &bat_priv->gw_list, list) {
> > +               if (((!gw_node->deleted) ||
> > +                    (time_after(jiffies, gw_node->deleted + timeout)))
> > && +                   atomic_read(&bat_priv->mesh_state) ==
> > MESH_ACTIVE) +                       continue;
> 
> This should be time_before.

Agreed.


> > -                             char *buff, size_t count)
> > +                            char *buff, size_t count)
> 
> Why was the space removed?

Should I move this into a separate patch ?


> > +               if ((strnicmp(tmp_ptr, "kbit", 4) == 0) ||
> > +                       (multi > 1))
> 
> Why this indentation? And This should be on a single line (or I count
> wrong).

No, you are right. Will be fixed.


> > +       if ((gw_mode_tmp == GW_MODE_CLIENT) &&
> > +           (atomic_read(&bat_priv->gw_sel_class) == 0))
> > +               atomic_set(&bat_priv->gw_sel_class, 20);
> > +
> > +       if ((gw_mode_tmp == GW_MODE_SERVER) &&
> > +           (atomic_read(&bat_priv->gw_bandwidth) == 0))
> > +               atomic_set(&bat_priv->gw_bandwidth, 41);
> 
> Why setting that here? And why do you set gw_sel_class here to 20 (if it is
> 0 before) and in softif_create to 0? Shouldn't it be set to 20 in
> softif_create instead?

The general idea was that if somebody activated client or gateway mode without 
setting gw_sel_class or gw_bandwidth to any value, it would use some default 
values. 
Moving these checks into softif_create() would only solve half of the problem 
because a user could set these variables to 0 and then start the client or 
server. At least the client mode requires the gw_sel_class variable value to 
be greater than 0.

Thanks for the comments,
Marek