Re: [PATCH v2] mmc: sdhci at91: add PM support

[Ludovic Desroches <ludovic.desroches@atmel.com>]

On Thu, Nov 26, 2015 at 05:24:40PM +0100, Ulf Hansson wrote:
> On 26 November 2015 at 17:07, Ludovic Desroches
> <ludovic.desroches@atmel.com> wrote:
> > On Tue, Nov 10, 2015 at 12:12:30PM +0100, Ulf Hansson wrote:
> >> On 10 November 2015 at 11:36, Ludovic Desroches
> >> <ludovic.desroches@atmel.com> wrote:
> >> > Add runtime PM support and use runtime_force_suspend|resume() for system
> >> > PM.
> >> >
> >
> > [...]
> >
> >> >  static int sdhci_at91_probe(struct platform_device *pdev)
> >> >  {
> >> >         const struct of_device_id       *match;
> >> > @@ -148,6 +206,11 @@ static int sdhci_at91_probe(struct platform_device *pdev)
> >> >         if (ret)
> >> >                 goto clocks_disable_unprepare;
> >> >
> >> > +       pm_runtime_set_active(&pdev->dev);
> >> > +       pm_runtime_enable(&pdev->dev);
> >> > +       pm_runtime_set_autosuspend_delay(&pdev->dev, 50);
> >> > +       pm_runtime_use_autosuspend(&pdev->dev);
> >>
> >> Move these four runtime PM calls above sdhci_add_host(), as after that
> >> point the host is used and thus runtime PM operations starts.
> >
> > Sadly I have discovered a bit late it is not working when doing this
> > because the controller has been suspended before doing the
> > sdhci_add_host().
> >
> > What is the right way to fix it? Calling pm_runtime_get_noresume()
> 
> I would do that before pm_runtime_enable().
> 
> > before sdhci_add_host() and calling pm_runtime_put_autosuspend() after?
> 
> Yes. Sorry for not spotting this before.

No problem, I am faulty too, it was not so trivial!

I am sending the fix. Thanks

Regards

Ludovic