[PATCH] kernel: DSA roaming fix for Marvell Link Street switch series

[Eneas U de Queiroz]

On Tue, Feb 2, 2021 at 6:15 AM DENG Qingfang <dqfext at gmail.com> wrote:
>
> Marvell Link Street switch series cannot perform MAC learning from
> CPU-injected (FROM_CPU) DSA frames, which results in 2 issues.
> - excessive flooding, due to the fact that DSA treats those addresses
> as unknown
> - the risk of stale routes, which can lead to temporary packet loss
>
> Backport those patch series from netdev mailing list, which solve these
> issues by adding and clearing static entries to the switch's FDB.
>
> Add a hack patch to set default VID to 1 in port_fdb_{add,del}. Otherwise
> the static entries will be added to the switch's private FDB if VLAN
> filtering disabled, which will not work.
>
> Link: https://lore.kernel.org/netdev/20210106095136.224739-1-olteanv@gmail.com/
> Link: https://lore.kernel.org/netdev/20210116012515.3152-1-tobias@waldekranz.com/
> Link: https://lore.kernel.org/netdev/20210130134334.10243-1-dqfext@gmail.com/
> Ref: https://gitlab.nic.cz/turris/turris-build/-/issues/165
> Signed-off-by: DENG Qingfang <dqfext at gmail.com>

Tested-by: Eneas U de Queiroz <cotequeiroz at gmail.com>

I have tested this using WRT3200ACM, and it solves the problem of
clients not able to roam from one AP to the another--my APs are wired,
not using WDS.  Clients would not be able to communicate for 300s
after roaming from one AP to another.  I consider this a critical bug,
so a fix must be included before 2021.02 branches.  I have applied the
patch to 3 APs, and have been using them for days without any real
issue--I'm not considering the 'ATU member violation' messages
reported earlier an issue, as they do appear to be harmless.

Cheers,

Eneas