On Wed, Jan 29, 2014 at 11:06:42AM +0100, Geert Uytterhoeven wrote:
> On Wed, Jan 29, 2014 at 7:17 AM, Simon Horman <horms@xxxxxxxxxxxx> wrote:
> >> >>+static struct of_device_id rcar_pci_of_match[] = {
> >> >>+ { .compatible = "renesas,pci-r8a7790", },
> >> >
> >> > Why only H2 SoC, if the driver is for both Gen2 SoCs?
> >>
> >> I can add a "renesas,pci-rcar-gen2" as a fallback match however
> >> I've not got anything other than an r8a7790 to test on. Also the
> >> compatible properties are preferably named after the soc.
> >
> > I'm not sure what the "best practice" is here but I propose that
> > you add both "renesas,pci-rcar-gen2" and "renesas,pci-r8a7790" now.
>
> To the driver or to the '90 DTS?
>
> If the DTS says
>
> compatible = "renesas,pci-r8a7790", "renesas,pci-rcar-gen2"
>
> then the driver can just live with "renesas,pci-rcar-gen2" for now.
>
> > And that "renesas,pci-r8a7791" can be added once the code has been
> > integrated and tested on that platform.
>
> And later we can add
>
> compatible = "renesas,pci-r8a7791", "renesas,pci-rcar-gen2"
>
> to the '91 DTS if it turns out to be compatible.
>
> If the '91 turns out to be incompatible, or partially compatible, the driver
> can be updated later to handle both "renesas,pci-r8a7790" and
> "renesas,pci-r8a7791" and differentiate between them.
>
> (That's what I was instructed to do with SPI ;-)
Thanks, that is more or less how I thought things work :)
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
