Em Qui, 2009-05-07 às 09:27 +0100, Mark Brown escreveu:
> On Thu, May 07, 2009 at 04:13:53AM -0300, Daniel Ribeiro wrote:
>
> > +#else
> > +static int pxamci_regulator_get(struct pxamci_host *host)
> > +{
> > + return 0;
> > +}
> > +static void pxamci_regulator_put(struct pxamci_host *host) {}
>
> regulator_get() and regulator_put() compile out if regulator is not
> defined, though you will get a dummy regulator back.
Will regulator_get() return an error case config_regulator is not
defined? Im relying on this to use the pdata case config_regulator is
not set.
>
> > - mmc->ocr_avail = host->pdata ?
> > + if (pxamci_regulator_get(&pdev->dev, host) == 0)
> > + mmc->ocr_avail = pxamci_regulator_get_ocrmask(host);
> > + else
> > + mmc->ocr_avail = host->pdata ?
> > host->pdata->ocr_mask :
> > MMC_VDD_32_33|MMC_VDD_33_34;
> > mmc->caps = 0;
>
> Are you sure this does the right thing if regulator API is compiled out?
> The dummy get() will succeed, meaning that we never take the platform
> data case which is probably not what you want.
No. This is wrong. I will send another version with this fixed.
>
> What should happen if the system both supplies platform data and a
> regulator?
If both are supplied and config_regulator is set, then the regulator
should be used. If config_regulator is not defined, or if the machine
does not supplies a regulator, then pdata->setpower and pdata->ocr_mask
should be used instead.
--
Daniel Ribeiro
