On Wed, Nov 14, 2018 at 10:37:31AM +0800, Hangbin Liu wrote:
> On Thu, Nov 08, 2018 at 03:44:10PM +0800, Hangbin Liu wrote:
> > On Fri, Oct 26, 2018 at 10:30:54AM +0800, Hangbin Liu wrote:
> > > Currently we only set mc_qrv to sysctl_mld_qrv when interface up. If we
> > > change sysctl_mld_qrv after interface up, it will has no effect.
> > >
> > > Fix it by assigning latest sysctl_mld_qrv to idev mc_qrv when join new
> > > group.
> > >
> > > Reported-by: Ying Xu <yi...@redhat.com>
> > > Signed-off-by: Hangbin Liu <liuhang...@gmail.com>
> >
> > Hi David,
> >
> > Any comments for this patch?
> > > ---
> > > net/ipv6/mcast.c | 1 +
> > > 1 file changed, 1 insertion(+)
> > >
> > > diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
> > > index dbab62e..bed4890 100644
> > > --- a/net/ipv6/mcast.c
> > > +++ b/net/ipv6/mcast.c
> > > @@ -680,6 +680,7 @@ static void igmp6_group_added(struct ifmcaddr6 *mc)
> > > if (!(dev->flags & IFF_UP) || (mc->mca_flags & MAF_NOREPORT))
> > > return;
> > >
> > > + mc->idev->mc_qrv = sysctl_mld_qrv;
> > > if (mld_in_v1_mode(mc->idev)) {
> > > igmp6_join_group(mc);
> > > return;
>
> Hi David,
>
> Sorry for the late reply. Today I find your reply in
> https://patchwork.ozlabs.org/patch/989422/, but I could not find your reply
> in my local mailbox, even on gmail web side. I can receive other kernel
> mails correctly, which is weird...I will check it later.
>
> For your question:
>
> > Why isn't mld_update_qrv() taking care of this?
>
> mld_update_qrv() is used for processing recieved mldv2 query message. This
> patch is to init new sysctl_mld_qrv for new joined groups, which haven't
> received any mld query message. e.g.
>
> - ipv6_mc_up() - ipv6_mc_reset() set idev->mc_qrv to default sysctl_mld_qrv,
> which is 2
> - Join group A with qrv = 2
> - user set sysctl_mld_qrv to 3
> - Join group B with qrv = 2 <- Here the new group should set qrv to 3 as
> user updated sysctl_mld_qrv. This is the
> issue this patch want to fix
> - Recived MLDv2 query message
> - Group A, B update new qrv via mld_update_qrv()
>
>
> > At a minimum, you must make your change take MLD_QRV_DEFAULT etc.
> > into account like mld_update_qrv() does.
>
> We should not init mc_qrv to min(MLD_QRV_DEFAULT, sysctl_mld_qrv), or
> user will never be able to set mc_qrv greater than 2.
>
Hi David,
Any comments? Should I re-send a new patch?
Thanks
Hangbin
