Re: [PATCH 2/8] mmc: sdhci: host: add new f_sdh30

[Arnd Bergmann <arnd@arndb.de>]

On Wednesday 16 July 2014 17:35:41 Vincent. Yang wrote:

> >
> >> +unsigned int sdhci_f_sdh30_get_min_clock(struct sdhci_host *host)
> >> +{
> >> +     return F_SDH30_MIN_CLOCK;
> >> +}
> >> +
> >> +void sdhci_f_sdh30_reset(struct sdhci_host *host, u8 mask)
> >> +{
> >> +     struct f_sdhost_priv *priv = sdhci_priv(host);
> >> +
> >> +     if (gpio_is_valid(priv->gpio_select_1v8))
> >> +             gpio_direction_output(priv->gpio_select_1v8, 1);
> >> +
> >> +     if (sdhci_readw(host, SDHCI_CLOCK_CONTROL) == 0) {
> >> +             sdhci_writew(host, 0xBC01, SDHCI_CLOCK_CONTROL);
> >> +             mmiowb();
> >> +     }
> >
> >Can you explain the mmiowb call here?
> 
> This came from the original 3.0 based driver. It's trying to be a
> write memory barrier. It wants to ensure the clock control change
> actually happened before the following code.
> I'll change it to a regular wmb in next version.

Note that with 'readw'/writew', you shouldn't need any extra barriers.
If sdhci_writew() is readw_relaxed(), wmb() is the correct barrier.

> >> +
> >> +     if (!of_property_read_u32(pdev->dev.of_node, "bus-width", &bus_width))
> >{
> >> +             switch (bus_width) {
> >> +             case 8:
> >> +                     dev_info(dev, "Applying 8 bit bus width\n");
> >> +                     host->mmc->caps |= MMC_CAP_8_BIT_DATA;
> >> +                     break;
> >> +             case 4:
> >> +                     dev_info(dev, "Applying 4 bit bus width\n");
> >> +                     host->mmc->caps |= MMC_CAP_4_BIT_DATA;
> >> +                     break;
> >> +             case 1:
> >> +             default:
> >> +                     dev_err(dev, "Invalid bus width: %u\n", bus_width);
> >> +                     break;
> >> +             }
> >> +     }
> >
> >This should probably be done in generic sdhci code somewhere. How about
> >adding it to sdhci_get_of_property instead_
> 
> I should use generic mmc_of_parse for it. I'll update it in next version.

Ah right, I thought we had this in common code already but couldn't find
it. Using mmc_of_parse() is definitely the correct solution here.

> >> +     priv->clk_sd4 = clk_get(&pdev->dev, "sd_sd4clk");
> >> +     if (!IS_ERR(priv->clk_sd4)) {
> >> +             ret = clk_prepare_enable(priv->clk_sd4);
> >> +             if (ret < 0) {
> >> +                     dev_err(dev, "Failed to enable sd4 clock: %d\n", ret);
> >> +                     goto err_clk1;
> >> +             }
> >> +     }
> >> +     priv->clk_b = clk_get(&pdev->dev, "sd_bclk");
> >> +     if (!IS_ERR(priv->clk_b)) {
> >> +             ret = clk_prepare_enable(priv->clk_b);
> >> +             if (ret < 0) {
> >> +                     dev_err(dev, "Failed to enable clk_b clock: %d\n", ret);
> >> +                     goto err_clk2;
> >> +             }
> >> +     }
> >
> >Please pick clock names that match what some of the other drivers use.
> 
> I'll use "iface" and "core" for clock names because they match what used
> in sdhci-msm driver.

Ok, makes sense.

	Arnd