[PATCH 5/5 v2] ARM: kirkwood: convert orion-wdt to fdt.
Jason
jason at lakedaemon.net
Sat Mar 3 06:57:10 EST 2012
On Fri, Mar 02, 2012 at 06:32:33PM +0000, Arnd Bergmann wrote:
> On Friday 02 March 2012, Jason Cooper wrote:
> > + wdt_reg = (void __iomem *)TIMER_VIRT_BASE;
> > +
> > + ret = of_address_to_resource(np, 0, &res);
> > + if (ret) {
> > + printk(KERN_ERR "invalid address\n");
> > + return ret;
> > + }
> > +
> > + wdt_reg = ioremap(res.start, res.end - res.start);
> > +
>
> This will break non-dt uses because you overwrite the wdt_reg value.
I thought for non-dt configurations of_.*() zeroed out. No problem, I
must've misunderstood something from earlier.
> As I commented before, I think the best approach would be to
> provide a hardcoded resource for the device in the legacy case
> and just use platform_get_resource() to get the resource
> and ioremap it in either case.
Sorry, my denseness got in the way. When you said hardcoded, I thought
you meant #define TIMER_VIRT_BASE .
> of_address_to_resource() works only for the DT case, while
> platform_get_resource() works in either case.
>
> You need to add something like
Ahh... now *that* makes sense.
> 8<---
> diff --git a/arch/arm/plat-orion/common.c b/arch/arm/plat-orion/common.c
> index 089899a..d16f43f 100644
> --- a/arch/arm/plat-orion/common.c
> +++ b/arch/arm/plat-orion/common.c
> @@ -568,13 +568,17 @@ void __init orion_spi_1_init(unsigned long mapbase,
> ****************************************************************************/
> static struct orion_wdt_platform_data orion_wdt_data;
>
> +static struct resource orion_wdt_resource =
> + DEFINE_RES_MEM(TIMER_PHYS_BASE, 0x28);
> +
> static struct platform_device orion_wdt_device = {
> .name = "orion_wdt",
> .id = -1,
> .dev = {
> .platform_data = &orion_wdt_data,
> },
> - .num_resources = 0,
> + .resources = &orion_wdt_resource,
> + .num_resources = 1,
> };
>
> void __init orion_wdt_init(unsigned long tclk)
> ---->8
>
> and an appropriate definition for TIMER_PHYS_BASE.
thx,
Jason.
More information about the devicetree-discuss
mailing list