Hi,
On Mon, 12 Apr 2021 at 06:58, Pavel Skripkin <[email protected]> wrote:
>
> syzbot reported WARNING in ieee802154_del_device. The problem
> was in uninitialized mutex. In case of NL802154_IFTYPE_MONITOR
> mutex won't be initialized, but ieee802154_del_device() accessing it.
>
> Reported-by: [email protected]
> Signed-off-by: Pavel Skripkin <[email protected]>
> ---
> net/mac802154/iface.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/net/mac802154/iface.c b/net/mac802154/iface.c
> index 1cf5ac09edcb..be8d2a02c882 100644
> --- a/net/mac802154/iface.c
> +++ b/net/mac802154/iface.c
> @@ -599,6 +599,7 @@ ieee802154_setup_sdata(struct ieee802154_sub_if_data *sdata,
>
> break;
> case NL802154_IFTYPE_MONITOR:
> + mutex_init(&sdata->sec_mtx);
> sdata->dev->needs_free_netdev = true;
> sdata->dev->netdev_ops = &mac802154_monitor_ops;
> wpan_dev->promiscuous_mode = true;
yes that will fix the issue, but will let the user notify that setting
any security setting is supported by monitors which is not the case.
There are patches around which should return -EOPNOTSUPP for monitors.
However we might support it in future to let the kernel encrypt air
frames, but this isn't supported yet and the user should be aware that
it isn't.
- Alex
Hi!
On Mon, 2021-04-12 at 07:45 -0400, Alexander Aring wrote:
> Hi,
>
> On Mon, 12 Apr 2021 at 06:58, Pavel Skripkin <[email protected]>
> wrote:
> >
> > syzbot reported WARNING in ieee802154_del_device. The problem
> > was in uninitialized mutex. In case of NL802154_IFTYPE_MONITOR
> > mutex won't be initialized, but ieee802154_del_device() accessing
> > it.
> >
> > Reported-by: [email protected]
> > Signed-off-by: Pavel Skripkin <[email protected]>
> > ---
> > net/mac802154/iface.c | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/net/mac802154/iface.c b/net/mac802154/iface.c
> > index 1cf5ac09edcb..be8d2a02c882 100644
> > --- a/net/mac802154/iface.c
> > +++ b/net/mac802154/iface.c
> > @@ -599,6 +599,7 @@ ieee802154_setup_sdata(struct
> > ieee802154_sub_if_data *sdata,
> >
> > break;
> > case NL802154_IFTYPE_MONITOR:
> > + mutex_init(&sdata->sec_mtx);
> > sdata->dev->needs_free_netdev = true;
> > sdata->dev->netdev_ops = &mac802154_monitor_ops;
> > wpan_dev->promiscuous_mode = true;
>
> yes that will fix the issue, but will let the user notify that
> setting
> any security setting is supported by monitors which is not the case.
> There are patches around which should return -EOPNOTSUPP for
> monitors.
> However we might support it in future to let the kernel encrypt air
> frames, but this isn't supported yet and the user should be aware
> that
> it isn't.
>
Thank you for your feedback. I am still not familiar with net internals
yet :) Next time I ll try to go deeper. Thanks!
> - Alex
With regards,
Pavel Skripkin