Re: [RFC/NOT FOR MERGING 1/3] arm: omap: use generic implementation if !od

[Kevin Hilman <khilman@linaro.org>]

Hi Felipe,

Felipe Balbi <balbi@ti.com> writes:

> Eventually, we need to be able to remove
> ti,hwmods DT attribute (or at a minimum
> ignore it).
>
> For new platforms, this patch could enable
> the transition by not relying on ti,hwmods
> to have functioning PM and Idle implementation.
>
> Notice that this poses no differences for
> platforms which are already supported, it
> just gives us means of dropping the relyance
> on hwmod for new platforms.
>
> NYET-Signed-off-by: Felipe Balbi <balbi@ti.com>
> ---
>  arch/arm/mach-omap2/omap_device.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
>
> diff --git a/arch/arm/mach-omap2/omap_device.c b/arch/arm/mach-omap2/omap_device.c
> index e065daa..305eeb4 100644
> --- a/arch/arm/mach-omap2/omap_device.c
> +++ b/arch/arm/mach-omap2/omap_device.c
> @@ -796,13 +796,18 @@ static int __init omap_early_device_register(struct platform_device *pdev)
>  static int _od_runtime_suspend(struct device *dev)
>  {
>  	struct platform_device *pdev = to_platform_device(dev);
> +	struct omap_device *od = to_omap_device(pdev);
>  	int ret;
>  
>  	ret = pm_generic_runtime_suspend(dev);
>  
> +	if (!od)
> +		goto out;
> +

Rather than adding a check for every function, I think you will get the
effect by simply not hooking up the PM domain.

IOW, in omap_device_build_from_dt(), conditionalize:

	pdev->dev.pm_domain = &omap_device_pm_domain;

then none of the callbacks will be called in the first place, so they
won't need to be conditionalized.

Without a PM domain, they will fallback to the bus level hooks, which in
this case will be the platform_bus hooks, which will just call the
pm_generic functions.

Kevin