On Thu, Feb 10, 2022 at 10:07:09AM +0100, Hans Schultz wrote:

On ons, feb 09, 2022 at 18:31, Ido Schimmel [off-list ref] wrote:

quoted

On Wed, Feb 09, 2022 at 02:05:32PM +0100, Hans Schultz wrote:

quoted

This series starts by adding support for SA filtering to the bridge,
which is then allowed to be offloaded to switchdev devices. Furthermore
an offloading implementation is supplied for the mv88e6xxx driver.

[...]

quoted

Hans Schultz (5):
  net: bridge: Add support for bridge port in locked mode
  net: bridge: Add support for offloading of locked port flag
  net: dsa: Add support for offloaded locked port flag
  net: dsa: mv88e6xxx: Add support for bridge port locked mode
  net: bridge: Refactor bridge port in locked mode to use jump labels

I think it is a bit weird to add a static key for this option when other
options (e.g., learning) don't use one. If you have data that proves
it's critical, then at least add it in patch #1 where the new option is
introduced.

Do you suggest that I drop patch #5 as I don't have data that it is
critical?

Yes. That way all the bridge options are consistent. In the future, if
someone reports a performance degradation (unlikely), all the options
can be converted to use a static key.

quoted

Please add a selftest under tools/testing/selftests/net/forwarding/. It
should allow you to test both the SW data path with veth pairs and the
offloaded data path with loopbacks. See tools/testing/selftests/net/forwarding/README

I will do that.

Thanks!