Re: [PATCH v6 2/4] mmc: sdhci-tegra: Add support to program MC stream ID

[Thierry Reding <thierry.reding@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 5104 bytes --]

On Wed, Sep 28, 2022 at 05:18:19PM +0300, Adrian Hunter wrote:
> On 28/09/22 15:56, Prathamesh Shete wrote:
> > SMMU clients are supposed to program stream ID from
> > their respective address spaces instead of MC override.
> > Define NVQUIRK_PROGRAM_STREAMID and use it to program
> > SMMU stream ID from the SDMMC client address space.
> > 
> > Signed-off-by: Aniruddha TVS Rao <anrao@nvidia.com>
> > Signed-off-by: Prathamesh Shete <pshete@nvidia.com>
> 
> One observation below.
> 
> Nevertheless:
> 
> Acked-by: Adrian Hunter <adrian.hunter@intel.com>
> 
> > ---
> >  drivers/mmc/host/sdhci-tegra.c | 44 ++++++++++++++++++++++++++++++++++
> >  1 file changed, 44 insertions(+)
> > 
> > diff --git a/drivers/mmc/host/sdhci-tegra.c b/drivers/mmc/host/sdhci-tegra.c
> > index a6c5bbae77b4..60ce3e80f248 100644
> > --- a/drivers/mmc/host/sdhci-tegra.c
> > +++ b/drivers/mmc/host/sdhci-tegra.c
> > @@ -25,6 +25,10 @@
> >  #include <linux/mmc/slot-gpio.h>
> >  #include <linux/gpio/consumer.h>
> >  #include <linux/ktime.h>
> > +#ifdef CONFIG_IOMMU_API
> > +#include <linux/iommu.h>
> > +#include <linux/bitops.h>
> > +#endif
> >  
> >  #include <soc/tegra/common.h>
> >  
> > @@ -94,6 +98,8 @@
> >  #define SDHCI_TEGRA_AUTO_CAL_STATUS			0x1ec
> >  #define SDHCI_TEGRA_AUTO_CAL_ACTIVE			BIT(31)
> >  
> > +#define SDHCI_TEGRA_CIF2AXI_CTRL_0			0x1fc
> > +
> >  #define NVQUIRK_FORCE_SDHCI_SPEC_200			BIT(0)
> >  #define NVQUIRK_ENABLE_BLOCK_GAP_DET			BIT(1)
> >  #define NVQUIRK_ENABLE_SDHCI_SPEC_300			BIT(2)
> > @@ -121,6 +127,7 @@
> >  #define NVQUIRK_HAS_TMCLK				BIT(10)
> >  
> >  #define NVQUIRK_HAS_ANDROID_GPT_SECTOR			BIT(11)
> > +#define NVQUIRK_PROGRAM_STREAMID			BIT(12)
> >  
> >  /* SDMMC CQE Base Address for Tegra Host Ver 4.1 and Higher */
> >  #define SDHCI_TEGRA_CQE_BASE_ADDR			0xF000
> > @@ -177,6 +184,9 @@ struct sdhci_tegra {
> >  	bool enable_hwcq;
> >  	unsigned long curr_clk_rate;
> >  	u8 tuned_tap_delay;
> > +#ifdef CONFIG_IOMMU_API
> > +	u32 streamid;
> > +#endif
> >  };
> >  
> >  static u16 tegra_sdhci_readw(struct sdhci_host *host, int reg)
> > @@ -1564,6 +1574,7 @@ static const struct sdhci_tegra_soc_data soc_data_tegra234 = {
> >  		    NVQUIRK_DIS_CARD_CLK_CONFIG_TAP |
> >  		    NVQUIRK_ENABLE_SDR50 |
> >  		    NVQUIRK_ENABLE_SDR104 |
> > +		    NVQUIRK_PROGRAM_STREAMID |
> >  		    NVQUIRK_HAS_TMCLK,
> >  	.min_tap_delay = 95,
> >  	.max_tap_delay = 111,
> > @@ -1775,6 +1786,25 @@ static int sdhci_tegra_probe(struct platform_device *pdev)
> >  	if (rc)
> >  		goto err_add_host;
> >  
> > +	/* Program MC streamID for DMA transfers */
> > +#ifdef CONFIG_IOMMU_API
> > +	if (soc_data->nvquirks & NVQUIRK_PROGRAM_STREAMID) {
> > +		struct iommu_fwspec *fwspec;
> > +
> > +		fwspec = dev_iommu_fwspec_get(&pdev->dev);
> > +		if (fwspec == NULL) {
> > +			dev_warn(mmc_dev(host->mmc),
> > +				"iommu fwspec is NULL, continue without stream ID\n");
> 
> It will still program a zero streamid upon resume.

This was confusing to me, but I think what you mean is that the check
for fwspec == NULL should also be done in sdhci_tegra_resume().

> 
> > +		} else {
> > +			tegra_host->streamid = fwspec->ids[0] & 0xff;
> > +			tegra_sdhci_writel(host, tegra_host->streamid |
> > +						FIELD_PREP(GENMASK(15, 8),
> > +						tegra_host->streamid),
> > +						SDHCI_TEGRA_CIF2AXI_CTRL_0);
> > +		}
> > +	}
> > +#endif
> > +
> >  	return 0;
> >  
> >  err_add_host:
> > @@ -1861,6 +1891,10 @@ static int sdhci_tegra_suspend(struct device *dev)
> >  static int sdhci_tegra_resume(struct device *dev)
> >  {
> >  	struct sdhci_host *host = dev_get_drvdata(dev);
> > +#ifdef CONFIG_IOMMU_API
> > +	struct sdhci_pltfm_host *pltfm_host = sdhci_priv(host);
> > +	struct sdhci_tegra *tegra_host = sdhci_pltfm_priv(pltfm_host);
> > +#endif
> >  	int ret;
> >  
> >  	ret = mmc_gpio_set_cd_wake(host->mmc, false);
> > @@ -1871,6 +1905,16 @@ static int sdhci_tegra_resume(struct device *dev)
> >  	if (ret)
> >  		return ret;
> >  
> > +	/* Re-program MC streamID for DMA transfers */
> > +#ifdef CONFIG_IOMMU_API
> > +	if (tegra_host->soc_data->nvquirks & NVQUIRK_PROGRAM_STREAMID) {
> > +		tegra_sdhci_writel(host, tegra_host->streamid |
> > +					FIELD_PREP(GENMASK(15, 8),
> > +					tegra_host->streamid),
> > +					SDHCI_TEGRA_CIF2AXI_CTRL_0);
> > +	}
> > +#endif

So this here should be exactly the same as in sdhci_tegra_probe(). Which
would be an argument for putting this into a separate function.

That said, I'm not sure if this matters at all. If fwspec == NULL, that
implies that device tree didn't have an "iommus" property, which also
implies that SMMU translations for this device will not be enabled at
all, if I recall correctly, so whether this CIF2AXI_CTRL register is
programmed or not in that case should be irrelevant.

Still, might be good to keep this the same in both sdhci_tegra_probe()
and sdhci_tegra_resume() for consistency.

With that fixed, the series:

Acked-by: Thierry Reding <treding@nvidia.com>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]