[PATCH] iommu: qcom-smmu: handle running in el2

[Neil Armstrong]

On 13/11/2024 06:09, Caleb Connolly wrote:
> We only need to configure the SMMU when running in EL1. In EL2 the
> hypervisor isn't running so peripherals can just do DMA as they wish.
> 
> Signed-off-by: Caleb Connolly <caleb.connolly at linaro.org>
> ---
>   drivers/iommu/qcom-hyp-smmu.c | 7 +++++++
>   1 file changed, 7 insertions(+)
> 
> diff --git a/drivers/iommu/qcom-hyp-smmu.c b/drivers/iommu/qcom-hyp-smmu.c
> index 1b5a09bb7b39..7b7b3ba01a07 100644
> --- a/drivers/iommu/qcom-hyp-smmu.c
> +++ b/drivers/iommu/qcom-hyp-smmu.c
> @@ -90,8 +90,10 @@ enum arm_smmu_s2cr_type {
>   struct qcom_smmu_priv {
>   	phys_addr_t base;
>   	struct list_head devices;
>   	struct udevice *dev;
> +	/* SMMU is not needed when running in EL2 */
> +	bool disable;
>   
>   	/* Read-once config */
>   	int num_cb;
>   	int num_smr;
> @@ -276,8 +278,11 @@ static int qcom_smmu_connect(struct udevice *dev)
>   	priv = dev_get_priv(dev->iommu);
>   	if (WARN_ON(!priv))
>   		return -EINVAL;
>   
> +	if (priv->disable)
> +		return 0;
> +
>   	mdev = alloc_dev(dev);
>   	if (IS_ERR(mdev) && PTR_ERR(mdev) != -EEXIST) {
>   		printf("%s: %s Couldn't create mmu context\n", __func__,
>   		       dev->name);
> @@ -347,8 +352,10 @@ static int qcom_smmu_probe(struct udevice *dev)
>   	priv->dev = dev;
>   	priv->base = dev_read_addr(dev);
>   	INIT_LIST_HEAD(&priv->devices);
>   
> +	priv->disable = current_el() > 1;
> +
>   	/* Read SMMU config */
>   	val = gr0_readl(priv, ARM_SMMU_GR0_ID0);
>   	priv->num_smr = FIELD_GET(ARM_SMMU_ID0_NUMSMRG, val);
>   

Reviewed-by: Neil Armstrong <neil.armstrong at linaro.org>