On Wed, Jul 16, 2025 at 12:15 PM Haiyang Zhang <haiya...@microsoft.com> wrote:
>
>
>
> > -----Original Message-----
> > From: Xin Long <lucien....@gmail.com>
> > Sent: Wednesday, July 16, 2025 11:14 AM
> > To: Simon Horman <ho...@kernel.org>
> > Cc: Li Tian <lit...@redhat.com>; net...@vger.kernel.org; linux-
> > hyp...@vger.kernel.org; linux-ker...@vger.kernel.org; Haiyang Zhang
> > <haiya...@microsoft.com>; Dexuan Cui <de...@microsoft.com>; Stephen
> > Hemminger <step...@networkplumber.org>; Long Li <lon...@microsoft.com>
> > Subject: [EXTERNAL] Re: [PATCH v3] hv_netvsc: Set VF priv_flags to
> > IFF_NO_ADDRCONF before open to prevent IPv6 addrconf
> >
> > On Wed, Jul 16, 2025 at 5:29 AM Simon Horman <ho...@kernel.org> wrote:
> > >
> > > + Xin Long
> > >
> > Thanks for Ccing me.
> >
> > > On Wed, Jul 16, 2025 at 08:26:05AM +0800, Li Tian wrote:
> > > > Set an additional flag IFF_NO_ADDRCONF to prevent ipv6 addrconf.
> > > >
> > > > Commit under Fixes added a new flag change that was not made
> > > > to hv_netvsc resulting in the VF being assinged an IPv6.
> > > >
> > > > Fixes: 8a321cf7becc ("net: add IFF_NO_ADDRCONF and use it in bonding
> > to prevent ipv6 addrconf")
> > > > Suggested-by: Cathy Avery <cav...@redhat.com>
> > > > Signed-off-by: Li Tian <lit...@redhat.com>
> > > > ---
> > > > v3:
> > > > - only fixes commit message.
> > > > v2:
> > https://lore.ker/
> > nel.org%2Fnetdev%2F20250710024603.10162-1-
> > litian%40redhat.com%2F&data=05%7C02%7Chaiyangz%40microsoft.com%7C80485948c
> > b344b12f2dd08ddc47b7c6e%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C63888
> > 2756868249313%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAu
> > MDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdat
> > a=1ljWIFtnhAiGjdIEgNNlQZGK%2F%2FZHdgHVkvyCWY9%2BKxI%3D&reserved=0
> > > > - instead of replacing flag, add it.
> > > > v1:
> > https://lore.ker/
> > nel.org%2Fnetdev%2F20250710024603.10162-1-
> > litian%40redhat.com%2F&data=05%7C02%7Chaiyangz%40microsoft.com%7C80485948c
> > b344b12f2dd08ddc47b7c6e%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C63888
> > 2756868272381%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAu
> > MDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdat
> > a=o%2B2BM9YEi3O2zcqQu9KfPae6PZerBWO%2FhL5KCIeJ9xI%3D&reserved=0
> > > > ---
> > > > drivers/net/hyperv/netvsc_drv.c | 5 ++++-
> > > > 1 file changed, 4 insertions(+), 1 deletion(-)
> > >
> > > Hi Li Tian,
> > >
> > > Thanks for addressing earlier feedback.
> > >
> > > I don't think you need to repost because of this, but for future
> > reference:
> > >
> > > 1. Because this is a fix for a commit that is present in net
> > > it should be targeted at that tree.
> > >
> > > Subject: [PATCH net vX] ...
> > >
> > > 2. Please use get_maintainers.pl this.patch to generate the CC list. In
> > > this case Xin Long (now CCed) should be included as he is the author
> > of the
> > > patch cited in the Fixes tag.
> > >
> > > b4 can help you with this and other aspects of patch management.
> > >
> > > >
> > > > diff --git a/drivers/net/hyperv/netvsc_drv.c
> > b/drivers/net/hyperv/netvsc_drv.c
> > > > index c41a025c66f0..8be9bce66a4e 100644
> > > > --- a/drivers/net/hyperv/netvsc_drv.c
> > > > +++ b/drivers/net/hyperv/netvsc_drv.c
> > > > @@ -2317,8 +2317,11 @@ static int netvsc_prepare_bonding(struct
> > net_device *vf_netdev)
> > > > if (!ndev)
> > > > return NOTIFY_DONE;
> > > >
> > > > - /* set slave flag before open to prevent IPv6 addrconf */
> > > > + /* Set slave flag and no addrconf flag before open
> > > > + * to prevent IPv6 addrconf.
> > > > + */
> > > > vf_netdev->flags |= IFF_SLAVE;
> > > > + vf_netdev->priv_flags |= IFF_NO_ADDRCONF;
> > If it is only to prevent IPv6 addrconf, I think you can replace IFF_SLAVE
> > with IFF_NO_ADDRCONF.
> >
> > IFF_SLAVE normally comes with IFF_MASTER, like bonding and eql.
> > I don't see IFF_MASTER used in netvsc_drv.c, so IFF_SLAVE probably
> > should be dropped, including the one in __netvsc_vf_setup()?
>
> The IFF_SLAVE is not just for ipv6, the comment should be updated.
> IFF_SLAVE is also used by udev and our other user mode daemons, so it needs
> to stay.
>
Got it, then the patch looks good to me.
Reviewed-by: Xin Long <lucien....@gmail.com>
