Re: [net-next PATCH v12 07/13] net: mdio: regmap: add support for multiple valid addr

[Christian Marangi <ansuelsmth@gmail.com>]

On Fri, Mar 14, 2025 at 09:01:56PM +0000, Russell King (Oracle) wrote:
> On Fri, Mar 14, 2025 at 08:41:33PM +0100, Andrew Lunn wrote:
> > On Sun, Mar 09, 2025 at 06:45:43PM +0100, Christian Marangi wrote:
> > > On Sun, Mar 09, 2025 at 05:36:49PM +0000, Russell King (Oracle) wrote:
> > > > On Sun, Mar 09, 2025 at 06:26:52PM +0100, Christian Marangi wrote:
> > > > > +/* If a non empty valid_addr_mask is passed, PHY address and
> > > > > + * read/write register are encoded in the regmap register
> > > > > + * by placing the register in the first 16 bits and the PHY address
> > > > > + * right after.
> > > > > + */
> > > > > +#define MDIO_REGMAP_PHY_ADDR		GENMASK(20, 16)
> > > > > +#define MDIO_REGMAP_PHY_REG		GENMASK(15, 0)
> > > > 
> > > > Clause 45 PHYs have 5 bits of PHY address, then 5 bits of mmd address,
> > > > and then 16 bits of register address - significant in that order. Can
> > > > we adjust the mask for the PHY address later to add the MMD between
> > > > the PHY address and register number?
> > > >
> > > 
> > > Honestly to future proof this, I think a good idea might be to add
> > > helper to encode these info and use Clause 45 format even for C22.
> > > Maybe we can use an extra bit to signal if the format is C22 or C45.
> > > 
> > > BIT(26) 0: C22 1:C45
> > > GENMASK(25, 21) PHY ADDR
> > > GENMASK(20, 16) MMD ADDR
> > > GENMASK(15, 0) REG
> > 
> > If you look back at older kernels, there was some helpers to do
> > something like this, but the C22/C45 was in bit 31. When i cleaned up
> > MDIO drivers to have separate C22 and C45 read/write functions, they
> > become redundant and they were removed. You might want to bring them
> > back again.
> 
> I'd prefer we didn't bring that abomination back. The detail about how
> things are stored in regmap should be internal within regmap, and I
> think it would be better to have an API presented that takes sensible
> parameters, rather than something that's been encoded.
>

Well problem is that regmap_write and regmap_read will take max 2 value
at the very end (reg and value) so it's really a matter of making the
encoding part internal but encoding it can't be skipped.

You are suggesting to introduce additional API like

mdio_regmap_write(regmap, phy, addr, val);
mdio_mmd_regmap_write(regmap, phy, mmd, addr, val);

And the encoding is done internally?

My concern is the decoding part from the .write/read_bits regmap OPs.
I guess for that also some helper should be exposed (to keep the
decoding/encoding internal to the driver and not expose the
_abomination_)

What do you think?

-- 
	Ansuel