On Tue, Nov 20, 2018 at 10:34 AM Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote:
> On Fri, 16 Nov 2018 at 00:23, Linus Walleij <linus.walleij@xxxxxxxxxx> wrote:
[Me]
> > Even if the slot GPIO core discards the inversion
> > semantics on the GPIO lines we take special care to
> > specify them right.
> > static struct s3c24xx_mci_pdata mini2440_mmc_cfg __initdata = {
> > - .gpio_detect = S3C2410_GPG(8),
> > - .gpio_wprotect = S3C2410_GPH(8),
> > - .wprotect_invert = 1,
>
> You missed to address one comment from my previous email - what about
> wprotect_invert? It seems it is gone after your changes.
Sorry about not adressing it.
So this is especially confusing. In platform data wprotect_invert is
specified, and in the driver this is handled like so:
if (pdata->wprotect_invert)
mmc->caps2 |= MMC_CAP2_RO_ACTIVE_HIGH;
Since the slot GPIO totally ignores the inversion flag on the gpio
descriptor (by reading the "raw" gpio value), we indeed need to keep
passing this as of now.
I am planning to get rid if the host caps for "RO_ACTIVE_HIGH"
etc at a later point.
I'll fix the patch.
Yours,
Linus Walleij
