Re: [PATCH 2/3] mmc: sdhci-of-at91: introduce driver for the Atmel SDMMC

[Ludovic Desroches <ludovic.desroches@atmel.com>]

On Sat, Jul 04, 2015 at 10:06:49PM +0200, Arnd Bergmann wrote:
> On Friday 03 July 2015 16:17:16 Ludovic Desroches wrote:
> > Introduce driver for he Atmel SDMMC available on sama5d2. It is a sdhci
> > compliant controller.
> > 
> > Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
> > ---
> >  .../devicetree/bindings/mmc/sdhci-atmel.txt        |  21 +++
> >  drivers/mmc/host/Kconfig                           |   8 +
> >  drivers/mmc/host/Makefile                          |   1 +
> >  drivers/mmc/host/sdhci-of-at91.c                   | 188 +++++++++++++++++++++
> >  4 files changed, 218 insertions(+)
> >  create mode 100644 Documentation/devicetree/bindings/mmc/sdhci-atmel.txt
> >  create mode 100644 drivers/mmc/host/sdhci-of-at91.c
> > 
> > diff --git a/Documentation/devicetree/bindings/mmc/sdhci-atmel.txt b/Documentation/devicetree/bindings/mmc/sdhci-atmel.txt
> > new file mode 100644
> > index 0000000..1b662d7
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/mmc/sdhci-atmel.txt
> > @@ -0,0 +1,21 @@
> > +* Atmel SDHCI controller
> > +
> > +This file documents the differences between the core properties in
> > +Documentation/devicetree/bindings/mmc/mmc.txt and the properties used by the
> > +sdhci-of-at91 driver.
> > +
> > +Required properties:
> > +- compatible:		Must be "atmel,sama5d2-sdhci".
> > +- clocks:		Phandlers to the clocks.
> > +- clock-names:		Must be "hclock", "multclk", "baseclk";
> > +
> 
> Are you sure that these are all new clocks that are unrelated to the "core",
> "mmc", and "iface" clocks that are used in other drivers?
> 

Maybe hclock could be considered as the core and interface clock. The
mmc clock will be baseclk or multclk. I think other devices manage the
multclk internally. Here we have to provide both clocks. Internally, we
only choose which clock to use and add a divider if necessary.

> > +
> > +	priv->mainck = devm_clk_get(&pdev->dev, "baseclk");
> > +	if (IS_ERR(priv->mainck)) {
> > +		dev_err(&pdev->dev, "failed to get baseclk\n");
> > +		return PTR_ERR(priv->mainck);
> > +	}
> > +
> > +	priv->hclock = devm_clk_get(&pdev->dev, "hclock");
> > +	if (IS_ERR(priv->hclock)) {
> > +		dev_err(&pdev->dev, "failed to get hclock\n");
> > +		return PTR_ERR(priv->hclock);
> > +	}
> > +
> > +	priv->gck = devm_clk_get(&pdev->dev, "multclk");
> > +	if (IS_ERR(priv->gck)) {
> > +		dev_err(&pdev->dev, "failed to get multclk\n");
> > +		return PTR_ERR(priv->gck);
> > +	}
> > +
> > +	host = sdhci_pltfm_init(pdev, soc_data, 0);
> > +	if (IS_ERR(host))
> > +		return PTR_ERR(host);
> > +
> > +	/*
> > +	 * The mult clock is provided by as a generated clock by the PMC
> > +	 * controller. In order to set the rate of gck, we have to get the
> > +	 * base clock rate and the clock mult from capabilities.
> > +	 */
> > +	clk_prepare_enable(priv->hclock);
> > +	caps0 = readl(host->ioaddr + SDHCI_CAPABILITIES);
> > +	caps1 = readl(host->ioaddr + SDHCI_CAPABILITIES_1);
> > +	clk_base = (caps0 & SDHCI_CLOCK_V3_BASE_MASK) >> SDHCI_CLOCK_BASE_SHIFT;
> > +	clk_mul = (caps1 & SDHCI_CLOCK_MUL_MASK) >> SDHCI_CLOCK_MUL_SHIFT;
> > +	gck_rate = clk_base * 1000000 * (clk_mul + 1);
> > +	ret = clk_set_rate(priv->gck, gck_rate);
> > +	if (ret < 0) {
> > +		dev_err(&pdev->dev, "failed to set gck");
> > +		goto hclock_disable_unprepare;
> > +		return -EINVAL;
> > +	}
> 
> Could this be shared with other drivers?

No, it's risk-free to do this.

> 
> 	Arnd