Re: [PATCH v4 net-next 5/5] net: mscc: ocelot: expose ocelot wm functions

On Tue, Dec 07, 2021 at 07:26:52AM -0800, Jakub Kicinski wrote:

On Tue, 7 Dec 2021 12:11:22 +0000 Vladimir Oltean wrote:

quoted

On Tue, Dec 07, 2021 at 11:48:52AM +0000, Russell King (Oracle) wrote:

quoted

Thank you for highlighting this.

Vladimir told me recently over the phylink get_interfaces vs get_caps
change for DSA, and I quote:

  David who applied your patch can correct me, but my understanding from
  the little time I've spent on netdev is that dead code isn't a candidate
  for getting accepted into the tree, even more so in the last few days
  before the merge window, from where it got into v5.16-rc1.
  ...
  So yes, I take issue with that as a matter of principle, I very much
  expect that a kernel developer of your experience does not set a
  precedent and a pretext for people who submit various shady stuff to the
  kernel just to make their downstream life easier.

This sounds very much like double-standards, especially as Vladimir
reviewed this.

I'm not going to be spiteful NAK these patches, because we all need to
get along with each other. I realise that it is sometimes useful to get
code merged that facilitates or aids further development - provided
that development is submitted in a timely manner.

I'm not taking this as a spiteful comment either, it is a very fair point.
Colin had previously submitted this as part of a 23-patch series and it
was me who suggested that this change could go in as part of preparation
work right away:
https://patchwork.kernel.org/project/netdevbpf/cover/20211116062328.1949151-1-colin.foster@in-advantage.com/#24596529
I didn't realize that in doing so with this particular change, we would
end up having some symbols exported by the ocelot switch lib that aren't
yet in use by other drivers. So yes, this would have to go in at the
same time as the driver submission itself.

I don't know the dependencies here (there are also pinctrl patches
in the linked series) so I'll defer to you, if there is a reason to
merge the unused symbols it needs to be spelled out, otherwise let's
drop the last patch for now.

I don't think there's any problem with dropping the last patch for now,
as that's the safer thing to do (Colin?), but just let us know whether
you prefer Colin to resend a 4-patch series, or you can pick this series
up without the last one.