Re: [PATCH] mmc: host: sdhci-esdhc-imx: add wakeup feature for GPIO CD pin

[Ulf Hansson <ulf.hansson@linaro.org>]

+ Rafael, linux-pm

On Tue, 26 May 2020 at 10:41, <haibo.chen@nxp.com> wrote:
>
> From: Haibo Chen <haibo.chen@nxp.com>
>
> When use the specific GPIO to detect the card insert/remove, we can
> also add the GPIO as a wakeup source. When system suspend, insert or
> remove the card can wakeup the system.

Yes, this makes perfect sense! However...

>
> Signed-off-by: Haibo Chen <haibo.chen@nxp.com>
> ---
>  drivers/mmc/host/sdhci-esdhc-imx.c | 19 +++++++++++++++++--
>  1 file changed, 17 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/mmc/host/sdhci-esdhc-imx.c b/drivers/mmc/host/sdhci-esdhc-imx.c
> index 5398af4824c3..7af9d87d4245 100644
> --- a/drivers/mmc/host/sdhci-esdhc-imx.c
> +++ b/drivers/mmc/host/sdhci-esdhc-imx.c
> @@ -1599,6 +1599,11 @@ static int sdhci_esdhc_imx_probe(struct platform_device *pdev)
>         if (esdhc_is_usdhc(imx_data)) {
>                 host->quirks2 |= SDHCI_QUIRK2_PRESET_VALUE_BROKEN;
>                 host->mmc->caps |= MMC_CAP_1_8V_DDR | MMC_CAP_3_3V_DDR;
> +
> +               /* GPIO CD can be set as a wakeup source */
> +               host->mmc->caps |= MMC_CAP_CD_WAKE;
> +               device_init_wakeup(&pdev->dev, true);
> +

... if the device is attached to a genpd (a PM domain), this may cause
it to stay powered on in the system suspend state. Not sure if that
can cause a problem for you? Wasting energy?

In any case, I think it's wrong to make "&pdev->dev" wakeup capable,
like this. Especially as at it's the GPIO controller that manages the
system wakeup irq.

>                 if (!(imx_data->socdata->flags & ESDHC_FLAG_HS200))
>                         host->quirks2 |= SDHCI_QUIRK2_BROKEN_HS200;
>
> @@ -1734,8 +1739,15 @@ static int sdhci_esdhc_suspend(struct device *dev)
>                 mmc_retune_needed(host->mmc);
>
>         ret = sdhci_suspend_host(host);
> -       if (!ret)
> -               return pinctrl_pm_select_sleep_state(dev);
> +       if (ret)
> +               return ret;
> +
> +       ret = pinctrl_pm_select_sleep_state(dev);
> +       if (ret)
> +               return ret;
> +
> +       if (device_may_wakeup(dev))
> +               ret = mmc_gpio_set_cd_wake(host->mmc, true);

With the above said, I think the best available approach at this
point, is to just call mmc_gpio_set_cd_wake() - and just skip checking
device_may_wakeup().

BTW, mmc_gpio_set_cd_wake() calls enable_irq_wake(), which doesn't
require the "dev" to be wakeup capable. So this is fine!

Yes, it means that the IRQ will be configured as a system wakeup,
without letting user space to have a say about it. I guess we can live
with that limitation.

>
>         return ret;
>  }
> @@ -1759,6 +1771,9 @@ static int sdhci_esdhc_resume(struct device *dev)
>         if (host->mmc->caps2 & MMC_CAP2_CQE)
>                 ret = cqhci_resume(host->mmc);
>
> +       if (!ret && device_may_wakeup(dev))
> +               ret = mmc_gpio_set_cd_wake(host->mmc, false);
> +
>         return ret;
>  }
>  #endif
> --
> 2.17.1
>

Kind regards
Uffe