On 2017/10/04 1:42, Ido Schimmel wrote: > On Tue, Oct 03, 2017 at 12:25:08PM -0400, Vivien Didelot wrote: >> Andrew Lunn <and...@lunn.ch> writes: >> >>>> The vlan will be effective only when vlan_filtering is enabled. >>>> When vlan_filtering is disabled, vlan information is still kept in the >>>> bridge and gets effective later when vlan_filtering becomes enable. >>> >>> O.K, so things are starting to get clearer. >>> >>> So when vlan filtering is disabled, the hardware should just ignore >>> the requests to add the vlan to the hardware? >>> >>> When vlan_filtering is enabled, are all the vlans in the software >>> bridge again offloaded? Or do we need to remember all the vlans which >>> we ignored while vlan filtering was disabled? The average switch has >>> nowhere to store these disabled vlans. It can only store active vlans.
Seems that __br_vlan_filter_toggle() only propagates SWITCHDEV_ATTR_ID_BRIDGE_VLAN_FILTERING. IMHO vlan-related objects (vlan, fdb, and mdb) should be remembered if vlan_filtering can be enabled later. But this sounds redundant as the same information is maintained in the bridge so I'm not sure this is the best way. >> >> When vlan_filtering is enabled on the bridge, the bridge code does >> propagates the default_pvid again if I recall correctly. I couldn't find it in the source... >> >> In my opinion the hardware mustn't ignore the VLAN requests, because we >> seem to agree that vlan_filtering disabled means that the target ports >> should not care yet about 802.1Q. So having some unused hardware VLAN >> entries and some ports with disabled 802.1Q mode must work together. Probably I don't fully understand you, but I think hardware can ignore VLAN requests while vlan_filtering is disabled, as long as they are properly populated to hardware on enabling vlan_filtering. >> >> That being said we still have the wrong hardware FDB populated when >> CONFIG_BRIDGE_VLAN_FILTERING is enabled but not vlan_filtering... > > The driver can make sure it's able to handle the configured > `vlan_filtering` state during port enslavement to the bridge and also > forbid it from being toggled once it's enslaved. That is a simple solution. One concern is backward compatibility. I wonder if we can prohibit toggling for some driver which currently allows it. -- Toshiaki Makita
