Re: [PATCH 2/2] net: gmii2rgmii: Switch priv field in mdio device structure

[Harini Katakam]

Hi Andrew,

On Tue, Aug 13, 2019 at 6:54 PM Andrew Lunn <andrew@xxxxxxx> wrote:
>
> On Tue, Aug 13, 2019 at 04:46:40PM +0530, Harini Katakam wrote:
> > Hi Andrew,
> >
> > On Thu, Aug 1, 2019 at 9:36 AM Andrew Lunn <andrew@xxxxxxx> wrote:
> > >
> > > On Wed, Jul 31, 2019 at 03:06:19PM +0530, Harini Katakam wrote:
> > > > Use the priv field in mdio device structure instead of the one in
> > > > phy device structure. The phy device priv field may be used by the
> > > > external phy driver and should not be overwritten.
> > >
> > > Hi Harini
> > >
> > > I _think_ you could use dev_set_drvdata(&mdiodev->dev) in xgmiitorgmii_probe() and
> > > dev_get_drvdata(&phydev->mdiomdio.dev) in _read_status()
> >
> > Thanks for the review. This works if I do:
> > dev_set_drvdata(&priv->phy_dev->mdio.dev->dev) in probe
> > and then
> > dev_get_drvdata(&phydev->mdio.dev) in _read_status()
> >
> > i.e mdiodev in gmii2rgmii probe and priv->phy_dev->mdio are not the same.
> >
> > If this is acceptable, I can send a v2.
>
> Hi Harini
>
> I think this is better, making use of the central driver
> infrastructure, rather than inventing something new.

Ok sure.

>
> The kernel does have a few helper, spi_get_drvdata, pci_get_drvdata,
> hci_get_drvdata. So maybe had add phydev_get_drvdata(struct phy_device
> *phydev)?

Maybe phydev_mdio_get_drvdata? Because the driver data member available is
phydev->mdio.dev.driver_data.

Regards,
Harini