Re: [PATCH 5/6] net: davinci_emac: Fix ioremap for devices with MDIO within the EMAC address space

[Felipe Balbi]

On Tue, Jan 13, 2015 at 11:59:58AM -0800, Tony Lindgren wrote:
> * Felipe Balbi <ba...@ti.com> [150113 11:57]:
> > On Tue, Jan 13, 2015 at 11:29:27AM -0800, Tony Lindgren wrote:
> > > Some devices like dm816x have the MDIO registers within the first EMAC
> > > instance address space. Let's fix the issue by allowing to pass an
> > > optional second IO range for the EMAC control register area.
> > > 
> > > Cc: Brian Hutchinson <b.hutch...@gmail.com>
> > > Cc: Felipe Balbi <ba...@ti.com>
> > > Signed-off-by: Tony Lindgren <t...@atomide.com>
> > > ---
> > >  drivers/net/ethernet/ti/davinci_emac.c | 15 ++++++++++++---
> > >  1 file changed, 12 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/drivers/net/ethernet/ti/davinci_emac.c 
> > > b/drivers/net/ethernet/ti/davinci_emac.c
> > > index 4c8d82c..0342273 100644
> > > --- a/drivers/net/ethernet/ti/davinci_emac.c
> > > +++ b/drivers/net/ethernet/ti/davinci_emac.c
> > > @@ -1877,7 +1877,7 @@ davinci_emac_of_get_pdata(struct platform_device 
> > > *pdev, struct emac_priv *priv)
> > >  static int davinci_emac_probe(struct platform_device *pdev)
> > >  {
> > >   int rc = 0;
> > > - struct resource *res;
> > > + struct resource *res, *res_ctrl;
> > >   struct net_device *ndev;
> > >   struct emac_priv *priv;
> > >   unsigned long hw_ram_addr;
> > > @@ -1936,11 +1936,20 @@ static int davinci_emac_probe(struct 
> > > platform_device *pdev)
> > >           rc = PTR_ERR(priv->remap_addr);
> > >           goto no_pdata;
> > >   }
> > > +
> > > + res_ctrl = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> > > + if (res_ctrl) {
> > 
> > devm_ioremap_resource() will check for res_ctrl being a valid pointer,
> > perhaps below would be slightly better ?
> > 
> > 
> >     res_ctrl = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> >     priv->ctrl_base = devm_ioremap_resource(&pdev->dev, res_ctrl);
> >     if (IS_ERR(priv->ctrl_base))
> >             priv->ctrl_base = priv->remap_addr + pdata->ctrl_mod_reg_offset;
> > 
> 
> We have a pile of devices using just one ioremap area so the second
> ioremap area needs to be optional. That's why we only do it based on
> if (res_ctrl).

fair enough

-- 
balbi

signature.asc
Description: Digital signature