[PATCH v6 13/25] iommu/arm-smmu-v3-kvm: Probe SMMU HW

[Jason Gunthorpe]

On Fri, May 01, 2026 at 11:19:15AM +0000, Mostafa Saleh wrote:
> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> index 61e6ab364086..157acde0436d 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> @@ -4738,12 +4738,6 @@ static int arm_smmu_device_reset(struct arm_smmu_device *smmu)
>  	return 0;
>  }
>  
> -#define IIDR_IMPLEMENTER_ARM		0x43b
> -#define IIDR_PRODUCTID_ARM_MMU_600	0x483
> -#define IIDR_PRODUCTID_ARM_MMU_700	0x487
> -#define IIDR_PRODUCTID_ARM_MMU_L1	0x48a
> -#define IIDR_PRODUCTID_ARM_MMU_S3	0x498
> -
>  static void arm_smmu_device_iidr_probe(struct arm_smmu_device *smmu)
>  {
>  	u32 reg;
> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> index 64618299d03a..f904f4d19609 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> @@ -84,6 +84,12 @@ struct arm_vsmmu;
>  #define IIDR_REVISION			GENMASK(15, 12)
>  #define IIDR_IMPLEMENTER		GENMASK(11, 0)
>  
> +#define IIDR_IMPLEMENTER_ARM		0x43b
> +#define IIDR_PRODUCTID_ARM_MMU_600	0x483
> +#define IIDR_PRODUCTID_ARM_MMU_700	0x487
> +#define IIDR_PRODUCTID_ARM_MMU_L1	0x48a
> +#define IIDR_PRODUCTID_ARM_MMU_S3	0x498
> +
>  #define ARM_SMMU_AIDR			0x1C

Lets put these hunks in some earlier patch to migrate out the
functions/etc

I think all these pkvm/arm-smmu-v3.c should just be building up the
driver.

> +static bool smmu_nesting_supported(struct hyp_arm_smmu_v3_device *smmu)
> +{
> +	unsigned int implementer, productid, variant, revision;
> +	u32 reg;
> +
> +	if (!(smmu->features & ARM_SMMU_FEAT_TRANS_S1) ||
> +	    !(smmu->features & ARM_SMMU_FEAT_TRANS_S2))
> +		return false;
> +
> +	reg = readl_relaxed(smmu->base + ARM_SMMU_IIDR);
> +	implementer = FIELD_GET(IIDR_IMPLEMENTER, reg);
> +	productid = FIELD_GET(IIDR_PRODUCTID, reg);
> +	variant = FIELD_GET(IIDR_VARIANT, reg);
> +	revision = FIELD_GET(IIDR_REVISION, reg);
> +
> +	if (implementer != IIDR_IMPLEMENTER_ARM)
> +		return true;
> +
> +	if (productid == IIDR_PRODUCTID_ARM_MMU_600)
> +		return variant >= 2;
> +	else if (productid == IIDR_PRODUCTID_ARM_MMU_700)
> +		return !(variant < 1 || revision < 1);
> +
> +	return true;
> +}

Why not share all this errata stuff with the idr parsing code too?

We already have ARM_SMMU_FEAT_NESTING that has the above calculation.

The two drivers use the same ARM_SMMU_FEAT system, I would expect one
chunk of shared code to compute the FEATs, who cares if pkvm doesn't
use all of them?

Use the same errata logic and so on to get to the feat bitmap.

Jason