On Fri, Oct 14, 2022 at 10:35:24AM +0000, Vladimir Oltean wrote:
> Hello Dan Carpenter,
>
> On Fri, Oct 14, 2022 at 12:41:05PM +0300, Dan Carpenter wrote:
> > Hello Vladimir Oltean,
> >
> > The patch acc43b7bf52a: "net: dsa: allow masters to join a LAG" from
> > Sep 11, 2022, leads to the following Smatch static checker warning:
> >
> > net/dsa/slave.c:3165 dsa_slave_netdevice_event()
> > error: uninitialized symbol 'err'.
> >
> > net/dsa/slave.c
> > 3145 case NETDEV_CHANGELOWERSTATE: {
> > 3146 struct netdev_notifier_changelowerstate_info *info = ptr;
> > 3147 struct dsa_port *dp;
> > 3148 int err;
> > 3149
> > 3150 if (dsa_slave_dev_check(dev)) {
> > 3151 dp = dsa_slave_to_port(dev);
> > 3152
> > 3153 err = dsa_port_lag_change(dp, info->lower_state_info);
> > 3154 }
> > 3155
> > 3156 /* Mirror LAG port events on DSA masters that are in
> > 3157 * a LAG towards their respective switch CPU ports
> > 3158 */
> > 3159 if (netdev_uses_dsa(dev)) {
> > 3160 dp = dev->dsa_ptr;
> > 3161
> > 3162 err = dsa_port_lag_change(dp, info->lower_state_info);
> > 3163 }
> >
> > Can both dsa_slave_dev_check() and netdev_uses_dsa() be false?
>
> Thanks for the report. Yes, it can happen that the logic falls through
> neither of those conditions. "err" needs to be initialized with 0 in
> that case. Could you please prepare a patch?
Yeah, I'll write a patch. Thanks!
regards,
dan carpenter
