Re: [RFC PATCH 05/11] media: iris: Skip DMA mask setting to core device when IOMMU is not mapped

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] Missing DMA segment size configuration for newly introduced subnodes.
--

--- Patch [5]: [RFC PATCH 05/11] media: iris: Skip DMA mask setting to core device when IOMMU is not mapped ---
commit 767f9e3fe023f5e91b49d0a9ac0e6e7678556f9b
Author: Vikash Garodia <vikash.garodia@oss.qualcomm.com>

media: iris: Skip DMA mask setting to core device when IOMMU is not mapped

This commit conditionally skips setting the DMA mask on the core device
if it does not have an IOMMU mapping. This addresses platforms where
subnodes are used for IOMMU streams instead of the parent device.

> diff --git a/drivers/media/platform/qcom/iris/iris_probe.c b/drivers/media/platform/qcom/iris/iris_probe.c
> index 0bce4fb1786bc..828e020fc79a7 100644
> --- a/drivers/media/platform/qcom/iris/iris_probe.c
> +++ b/drivers/media/platform/qcom/iris/iris_probe.c
> @@ -309,9 +309,11 @@ static int iris_probe(struct platform_device *pdev)
>  
>  	dma_mask = core->iris_platform_data->dma_mask;
>  
> -	ret = dma_set_mask_and_coherent(dev, dma_mask);
> -	if (ret)
> -		goto err_vdev_unreg_enc;
> +	if (device_iommu_mapped(dev)) {
> +		ret = dma_set_mask_and_coherent(dev, dma_mask);
> +		if (ret)
> +			goto err_vdev_unreg_enc;
> +	}
>  
>  	dma_set_max_seg_size(&pdev->dev, DMA_BIT_MASK(32));
>  	dma_set_seg_boundary(&pdev->dev, DMA_BIT_MASK(32));

[Severity: Low]
Since the DMA mask is conditionally skipped for the parent device when
subnodes are in use, do we also need to set these max segment size and
boundary limits on the subnode devices that are actually performing the
DMA?

If an application imports a large dma-buf for video streaming, the subnode
devices appear to inherit the default 64KB maximum segment size limit. When
vb2_dma_contig() calls dma_map_sgtable() using the subnode device, the IOMMU
driver might attempt to process segments larger than the unconfigured 64KB
default.

Could this trigger a DMA API debug warning ("mapping sg segment longer
than device claims to support") and introduce unnecessary CPU overhead
as the IOMMU driver evaluates splitting the mapping?

Would it make sense to configure these DMA parameters for the subnodes
in iris_create_cb_dev() as well?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260709-vpu_iommu_iova_handling-v1-0-72bb62cb2dfd@oss.qualcomm.com?part=5