On 07/03/2022 04:34, Zhenxiong Lai wrote:
> From: Zhenxiong Lai <zhenxiong.lai@xxxxxxxxxx>
>
> DLL locked status probably couldn't come out in a certain time on
> Sharkl3 platform sometimes.
> It can be workaround if re-anable card en bit only
> rather than calling ->set_clock() after UHS timing changed.
>
> Signed-off-by: Zhenxiong Lai <zhenxiong.lai@xxxxxxxxxx>
> ---
> drivers/mmc/host/sdhci.c | 7 ++++++-
> drivers/mmc/host/sdhci.h | 2 ++
This is an SDHCI patch so it would be better if the subject
begins with "mmc: sdhci: "
> 2 files changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
> index 07c6da1f2f0f..c415d00304fe 100644
> --- a/drivers/mmc/host/sdhci.c
> +++ b/drivers/mmc/host/sdhci.c
> @@ -2396,7 +2396,12 @@ void sdhci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios)
> }
>
> /* Re-enable SD Clock */
> - host->ops->set_clock(host, host->clock);
> + if (host->quirks2 & SDHCI_QUIRK2_RE_ENABLE_CARD_EN) {
> + clk = sdhci_readw(host, SDHCI_CLOCK_CONTROL);
> + clk |= SDHCI_CLOCK_CARD_EN;
> + sdhci_writew(host, clk, SDHCI_CLOCK_CONTROL);
We like to try and avoid quirks whenever possible.
Why can't this be done in the driver's ->set_clock() callback?
> + } else
> + host->ops->set_clock(host, host->clock);
> } else
> sdhci_writeb(host, ctrl, SDHCI_HOST_CONTROL);
>
> diff --git a/drivers/mmc/host/sdhci.h b/drivers/mmc/host/sdhci.h
> index d7929d725730..3d4ab68f133f 100644
> --- a/drivers/mmc/host/sdhci.h
> +++ b/drivers/mmc/host/sdhci.h
> @@ -477,6 +477,8 @@ struct sdhci_host {
> * block count.
> */
> #define SDHCI_QUIRK2_USE_32BIT_BLK_CNT (1<<18)
> +/* controller only need to enable card en again after UHS timing was changed */
> +#define SDHCI_QUIRK2_RE_ENABLE_CARD_EN (1<<19)
A quirk without any users is not useful. Please also send the driver
change that uses this quirk, although as I wrote above, the ->set_clock()
callback should be used instead.
>
> int irq; /* Device IRQ */
> void __iomem *ioaddr; /* Mapped address */
