Re: [PATCH v2 net-next 08/14] net: dsa: felix: update init_regmap to be string-based

[Colin Foster <colin.foster@xxxxxxxxxxxxxxxx>]

On Thu, Sep 22, 2022 at 07:40:09PM -0700, Jakub Kicinski wrote:
> On Thu, 22 Sep 2022 19:39:06 -0700 Jakub Kicinski wrote:
> > On Wed, 21 Sep 2022 21:00:56 -0700 Colin Foster wrote:
> > > During development, it was believed that a wrapper for ocelot_regmap_init()
> > > would be sufficient for the felix driver to work in non-mmio scenarios.
> > > This was merged in during commit 242bd0c10bbd ("net: dsa: ocelot: felix:
> > > add interface for custom regmaps")
> > > 
> > > As the external ocelot DSA driver grew closer to an acceptable state, it
> > > was realized that most of the parameters that were passed in from struct
> > > resource *res were useless and ignored. This is due to the fact that the
> > > external ocelot DSA driver utilizes dev_get_regmap(dev, resource->name).
> > > 
> > > Instead of simply ignoring those parameters, refactor the API to only
> > > require the name as an argument. MMIO scenarios this will reconstruct the
> > > struct resource before calling ocelot_regmap_init(ocelot, resource). MFD
> > > scenarios need only call dev_get_regmap(dev, name).
> 
> Ah, and the modpost:
> 
> ERROR: modpost: drivers/net/dsa/ocelot/mscc_seville: 'felix_init_regmap' exported twice. Previous export was in drivers/net/dsa/ocelot/mscc_felix.ko

Yep. Since felix.c gets compiled into both drivers I don't need to
export the symbol.

And there's the NULL assignment. That one I'm surprised doesn't throw a
warning... Looking forward to more feedback before I send out v3 next
week.