Re: [net-next RFC PATCH 1/4] net: dsa: qca8k: drop qca8k_read/write/rmw for regmap variant

[Vladimir Oltean]

On Tue, Jul 19, 2022 at 02:17:20AM +0200, Christian Marangi wrote:
> Wonder if a good idea would be leave things as is for now and work of a
> single dsa_switch_ops on another series.
> 
> With "leave things as is" I mean that function will get migrated to
> qca8k-common.c and exposed with the header file.
> 
> And the dsa_switch_ops is defined in qca8k specific code.
> 
> The warn about the 23 patch was scary so considering this series is
> already a bit big and I can squash only a few patch, putting extra logic
> to correctly handle each would make this even bigger.
> 
> Think the right thing to do is handling the changes for single
> dsa_switch_ops to a separate series and at the same time also get some
> info on ipq4019 and what can be generalized.
> 
> What do you think?

I don't have a clear mental image right now of how things would look like,
but I suppose you can try and I can review the result. I imagine the
only code added now that you'll need to delete when you later migrate from
switch-specific dsa_switch_ops to common dsa_switch_ops are the function
prototypes from qca8k.h, since the implementations of the dsa_switch_ops
will become static functions at some point in the future.