> Subject: Re: [PATCH] hv_netvsc: Mark VF as slave before exposing it to user-mode
>
> On Wed, 25 Oct 2023 15:50:50 -0700
> longli@xxxxxxxxxxxxxxxxx wrote:
>
> > @@ -2347,6 +2342,12 @@ static int netvsc_register_vf(struct net_device
> *vf_netdev)
> > if (!ndev)
> > return NOTIFY_DONE;
> >
> > + if (event == NETDEV_POST_INIT) {
> > + /* set slave flag before open to prevent IPv6 addrconf */
> > + vf_netdev->flags |= IFF_SLAVE;
> > + return NOTIFY_DONE;
> > + }
> > +
> > net_device_ctx = netdev_priv(ndev);
> > netvsc_dev = rtnl_dereference(net_device_ctx->nvdev);
> > if (!netvsc_dev || rtnl_dereference(net_device_ctx->vf_netdev))
> > @@ -2753,8 +2754,9 @@ static int netvsc_netdev_event(struct notifier_block
> *this,
> > return NOTIFY_DONE;
> >
> > switch (event) {
> > + case NETDEV_POST_INIT:
> > case NETDEV_REGISTER:
> > - return netvsc_register_vf(event_dev);
> > + return netvsc_register_vf(event_dev, event);
>
> Although correct, this is an awkward way to write this.
> There are two events which call register_vf() but the post init one short circuits
> and doesn't really register the VF.
>
> The code is clearer if flag is set in switch statement.
>
I will add a dedicated function to handle NETDEV_POST_INIT.
Thanks,
Long
