Re: [PATCH net] net: bridge: Do not offload IGMP/MLD messages

[Joseph Huang <joseph.huang.2024@gmail.com>]

On 7/7/2025 9:00 AM, Vladimir Oltean wrote:
> Hi Joseph,
> 
> On Tue, Jul 01, 2025 at 03:36:38PM -0400, Joseph Huang wrote:
>> Do not offload IGMP/MLD messages as it could lead to IGMP/MLD Reports
>> being unintentionally flooded to Hosts. Instead, let the bridge decide
>> where to send these IGMP/MLD messages.
>> 
>> Fixes: 472111920f1c ("net: bridge: switchdev: allow the TX data plane forwarding to be offloaded")
>> Signed-off-by: Joseph Huang <Joseph.Huang@garmin.com>
>> ---
>>  net/bridge/br_switchdev.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>> 
>> diff --git a/net/bridge/br_switchdev.c b/net/bridge/br_switchdev.c
>> index 95d7355a0407..757c34bf5931 100644
>> --- a/net/bridge/br_switchdev.c
>> +++ b/net/bridge/br_switchdev.c
>> @@ -18,7 +18,8 @@ static bool nbp_switchdev_can_offload_tx_fwd(const struct net_bridge_port *p,
>>                 return false;
>> 
>>         return (p->flags & BR_TX_FWD_OFFLOAD) &&
>> -              (p->hwdom != BR_INPUT_SKB_CB(skb)->src_hwdom);
>> +              (p->hwdom != BR_INPUT_SKB_CB(skb)->src_hwdom) &&
>> +              !br_multicast_igmp_type(skb);
>>  }
>> 
>>  bool br_switchdev_frame_uses_tx_fwd_offload(struct sk_buff *skb)
>> --
>> 2.49.0
>>
> 
> Can you please incorporate the extra clarifications made to Tobias in
> the commit message? They provide valuable background.

Will do.

> 
> Also, keeping in mind I have no experience with IGMP/MLD snooping:
> aren't there IGMP/MLD messages which should be delivered to all hosts?
> Are you looking for BR_INPUT_SKB_CB_MROUTERS_ONLY(skb) as a substitute
> to br_multicast_igmp_type() instead?
> 

Yes, there are messages which should be delivered to all hosts, but if 
we offload these messages, we're at the mercy of the multicast_flood 
setting on each port and whether or not 224.0.0.1/ff02::1 are "known" 
multicast groups. (There's a discrepancy between the bridge and DSA 
subsystem on how 224.0.0.1/ff02::1 groups are handled.)

Group Specific Queries might be the only ones which are safe to offload 
regardless of what multicast_flood setting is, but I don't think the 
bridge has a flag specifically for those.

Thanks,
Joseph