On Wed, Mar 16, 2022 at 10:51:17AM +0100, Tobias Waldekranz wrote:

quoted

Question though:

quoted

quoted

+			err = dsa_port_msti_fast_age(dp, state->msti);

If _msti_fast_age returns an error here, do we want that to bubble up to
the bridge? It seems more important to keep the bridge in sync with the
hardware. I.e. the hardware state has already been successfully synced,
we just weren't able to flush all VLANs for some reason. We could revert
the state I guess, but what if that fails?

Should we settle for a log message?

Or should we set the extack message? Similar to how we report software
fallback of bridging/LAGs?

A warning extack and chug along sounds great. The worst that can happen
if flushing a VLAN's FDB fails is that the topology will reconverge slower.