EAP AP/VLAN: multicast not send to client

[Sebastian Gottschall]

>> then it might be something different. i'm not sure at that point. but
>> consider that AP_VLAN is also used for WDS STA/WDS AP implementations.
>> so each wds station becomes a individual interface on ap side for
>> bridging. that might of course have a influence to broadcast/multicast
>> frames
> Interestingly, the WDS "mode" works perfectly fine  on the same hardware. So I
> have not a real idea what the difference is between EAP AP_VLAN and WDS (PSK)
> AP_VLAN at the moment.

mmh. l have a idea

try the following (this a patch in my tree) and check also the wmi 
services for this service flag which might be a difference between these 
firmwares

--- a/drivers/net/wireless/ath/ath10k/mac.c
+++ b/drivers/net/wireless/ath/ath10k/mac.c
@@ -9003,10 +9003,10 @@ int ath10k_mac_register(struct ath10k *ar)
                goto err_dfs_detector_exit;
        }

-       if (test_bit(WMI_SERVICE_PER_PACKET_SW_ENCRYPT, 
ar->wmi.svc_map)) {
+//     if (test_bit(WMI_SERVICE_PER_PACKET_SW_ENCRYPT, 
ar->wmi.svc_map)) {
                ar->hw->wiphy->interface_modes |= 
BIT(NL80211_IFTYPE_AP_VLAN);
                ar->hw->wiphy->software_iftypes |= 
BIT(NL80211_IFTYPE_AP_VLAN);
-       }
+//     }

        if (!ath_is_world_regd(&ar->ath_common.regulatory)) {
                ret = regulatory_hint(ar->hw->wiphy,

>
>> in addition the 10.4 firmware series has a special wmi flag which is not
>> used by ath10k, but solved alot of problems

>>   for me (in my case with
>> reauthentication)
>> WMI_VDEV_DISABLE_4_ADDR_SRC_LRN
>> this might be unrelated to your issue. but this flag must be set anyway
>> for proper work. but it isnt by default in all 10.4 firmwares.
> Good to know, thanks.
>
>>> For Ben Greear's firmware, the only workaround which I found until now was to
>>> set NL80211_ATTR_MULTICAST_TO_UNICAST_ENABLED to 1 to force mac80211 to send
>>> multicast as unicast.
>> so if its a pure multicast/unicast issue it could be also a basic rate
>> issue for ct firmwares.
> Interesting idea. Btw. I've also switched to sw crypto for a test just to make
> sure that there is no HW key problem.
>
> Kind regards,
> 	Sven