On Mon, Nov 18, 2013 at 02:25:40PM +0100, Stefan Hajnoczi wrote:
> On Mon, Nov 18, 2013 at 04:20:12PM +0800, Amos Kong wrote:
> > @@ -967,6 +968,27 @@ void print_net_client(Monitor *mon, NetClientState *nc)
> > nc->info_str);
> > }
> >
> > +void rxfilter_notify(NetClientState *nc, Object *obj)
> > +{
> > + QObject *event_data;
> > + gchar *path = object_get_canonical_path(obj);
> > +
> > + if (nc->rxfilter_notify_enabled) {
> > + if (nc->name) {
> > + event_data = qobject_from_jsonf("{ 'name': %s, 'path': %s }",
> > + nc->name, path);
> > + } else {
> > + event_data = qobject_from_jsonf("{ 'path': %s }", path);
> > + }
> > + monitor_protocol_event(QEVENT_NIC_RX_FILTER_CHANGED, event_data);
> > + qobject_decref(event_data);
> > +
> > + /* disable event notification to avoid events flooding */
> > + nc->rxfilter_notify_enabled = 0;
>
> Only hw/net/virtio-net.c uses nc->rxfilter_notify_enabled. This
> function isn't reusable in its current form so I'm left wondering what
> the point of this patch is?
>
> If you have patches that invoke rxfilter_notify() from other NICs then
> please submit them together in a series.
I don't have patch to change other NICs to send rxfilter notify right
now.
So I will send a patch to fix the leak. We can move the function to
net.c in future.
> Otherwise, let's not move things around just for the sake of it,
> especially when the refactoring is not done correctly.
>
> Stefan
--
Amos.
