Re: [PATCH RESEND v3 2/3] net: mdio: add reset control for Aspeed MDIO

[Philipp Zabel]

On Mo, 2022-04-11 at 09:50 +0930, Andrew Jeffery wrote:
> 
> On Thu, 7 Apr 2022, at 17:27, Dylan Hung wrote:
> > Add reset assertion/deassertion for Aspeed MDIO.  There are 4 MDIO
> > controllers embedded in Aspeed AST2600 SOC and share one reset control
> > register SCU50[3].  To work with old DT blobs which don't have the reset
> > property, devm_reset_control_get_optional_shared is used in this change.
> > 
> > Signed-off-by: Dylan Hung <dylan_hung@xxxxxxxxxxxxxx>
> > Reviewed-by: Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>
> > ---
> >  drivers/net/mdio/mdio-aspeed.c | 15 ++++++++++++++-
> >  1 file changed, 14 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/net/mdio/mdio-aspeed.c b/drivers/net/mdio/mdio-aspeed.c
> > index e2273588c75b..1afb58ccc524 100644
> > --- a/drivers/net/mdio/mdio-aspeed.c
> > +++ b/drivers/net/mdio/mdio-aspeed.c
> > @@ -3,6 +3,7 @@
> > 
> >  #include <linux/bitfield.h>
> >  #include <linux/delay.h>
> > +#include <linux/reset.h>
> >  #include <linux/iopoll.h>
> >  #include <linux/mdio.h>
> >  #include <linux/module.h>
> > @@ -37,6 +38,7 @@
> > 
> >  struct aspeed_mdio {
> >  	void __iomem *base;
> > +	struct reset_control *reset;
> >  };
> > 
> >  static int aspeed_mdio_read(struct mii_bus *bus, int addr, int regnum)
> > @@ -120,6 +122,12 @@ static int aspeed_mdio_probe(struct platform_device *pdev)
> >  	if (IS_ERR(ctx->base))
> >  		return PTR_ERR(ctx->base);
> > 
> > +	ctx->reset = devm_reset_control_get_optional_shared(&pdev->dev, NULL);
> > +	if (IS_ERR(ctx->reset))
> > +		return PTR_ERR(ctx->reset);
> > +
> > +	reset_control_deassert(ctx->reset);
> > +
> >  	bus->name = DRV_NAME;
> >  	snprintf(bus->id, MII_BUS_ID_SIZE, "%s%d", pdev->name, pdev->id);
> >  	bus->parent = &pdev->dev;
> > @@ -129,6 +137,7 @@ static int aspeed_mdio_probe(struct platform_device *pdev)
> >  	rc = of_mdiobus_register(bus, pdev->dev.of_node);
> >  	if (rc) {
> >  		dev_err(&pdev->dev, "Cannot register MDIO bus!\n");
> > +		reset_control_assert(ctx->reset);
> >  		return rc;
> >  	}
> > 
> > @@ -139,7 +148,11 @@ static int aspeed_mdio_probe(struct platform_device *pdev)
> > 
> >  static int aspeed_mdio_remove(struct platform_device *pdev)
> >  {
> > -	mdiobus_unregister(platform_get_drvdata(pdev));
> > +	struct mii_bus *bus = (struct mii_bus *)platform_get_drvdata(pdev);
> > +	struct aspeed_mdio *ctx = bus->priv;
> > +
> > +	reset_control_assert(ctx->reset);
> 
> Isn't this unnecessary because you've used the devm_ variant to acquire 
> the reset?

No, this is correct. deassert/assert needs to be balanced, and the
reset_control_deassert() call in aspeed_mdio_probe() is not devres
managed.

regards
Philipp