[net-next,v3,2/3] bridge: only expire the mdb entry when query is received

From: Cong Wang <amwang@redhat.com>

Currently we arm the expire timer when the mdb entry is added,
however, this causes problem when there is no querier sent
out after that.

So we should only arm the timer when a corresponding query is
received, as suggested by Herbert.

And "if there is no querier then group subscriptions shouldn't expire.
There has to be at least one querier in the network for this thing to
work.  Otherwise it just degenerates into a non-snooping switch,
which is OK."

Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Adam Baker <linux@baker-net.org.uk>
Signed-off-by: Cong Wang <amwang@redhat.com>
---
v3: fix the code in br_multicast_leave_group()
v2: add a flag to check if timer is fired or not

 net/bridge/br_multicast.c |   21 ++++++++++++---------
 net/bridge/br_private.h   |    1 +
 2 files changed, 13 insertions(+), 9 deletions(-)

On Tue, Apr 30, 2013 at 03:10:18PM +0800, Cong Wang wrote:
>
> @@ -1275,7 +1277,7 @@ static void br_multicast_leave_group(struct net_bridge *br,
>  		     br->multicast_last_member_interval;
>  
>  	if (!port) {
> -		if (mp->mglist &&
> +		if (mp->mglist && mp->timer_armed &&
>  		    (timer_pending(&mp->timer) ?
>  		     time_after(mp->timer.expires, time) :
>  		     try_to_del_timer_sync(&mp->timer) >= 0)) {

This whole section of code should now be removed since we're now
only arming the timer in response to a query to that group.  Of
course if we were the querier and sent out a query here then you'd
need to simulate the reception of a query.

Cheers,

On Thu, 2013-05-02 at 09:59 +0800, Herbert Xu wrote:
> On Tue, Apr 30, 2013 at 03:10:18PM +0800, Cong Wang wrote:
> >
> > @@ -1275,7 +1277,7 @@ static void br_multicast_leave_group(struct net_bridge *br,
> >  		     br->multicast_last_member_interval;
> >  
> >  	if (!port) {
> > -		if (mp->mglist &&
> > +		if (mp->mglist && mp->timer_armed &&
> >  		    (timer_pending(&mp->timer) ?
> >  		     time_after(mp->timer.expires, time) :
> >  		     try_to_del_timer_sync(&mp->timer) >= 0)) {
> 
> This whole section of code should now be removed since we're now
> only arming the timer in response to a query to that group.  Of
> course if we were the querier and sent out a query here then you'd
> need to simulate the reception of a query.

Hmm, this piece of code is checking if bridge itself is in the mdb entry
and if it just receives a leave, which seems still necessary... What am
I missing?

Thanks!

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On Fri, May 03, 2013 at 11:50:22AM +0800, Cong Wang wrote:
> On Thu, 2013-05-02 at 09:59 +0800, Herbert Xu wrote:
> > On Tue, Apr 30, 2013 at 03:10:18PM +0800, Cong Wang wrote:
> > >
> > > @@ -1275,7 +1277,7 @@ static void br_multicast_leave_group(struct net_bridge *br,
> > >  		     br->multicast_last_member_interval;
> > >  
> > >  	if (!port) {
> > > -		if (mp->mglist &&
> > > +		if (mp->mglist && mp->timer_armed &&
> > >  		    (timer_pending(&mp->timer) ?
> > >  		     time_after(mp->timer.expires, time) :
> > >  		     try_to_del_timer_sync(&mp->timer) >= 0)) {
> > 
> > This whole section of code should now be removed since we're now
> > only arming the timer in response to a query to that group.  Of
> > course if we were the querier and sent out a query here then you'd
> > need to simulate the reception of a query.
> 
> Hmm, this piece of code is checking if bridge itself is in the mdb entry
> and if it just receives a leave, which seems still necessary... What am
> I missing?

I think I quoted the wrong hunk in the patch, I meant the code
that arms the timer should no longer be in the leave_group function
unless we just sent a query ourselves (and in that case the expiration
should also be adjusted accordingly).

Cheers,