diff mbox series

iommu/dma: Document min_align_mask assumption

Message ID dbb4d2d8e5d1691ac9a6c67e9758904e6c447ba5.1709553942.git.robin.murphy@arm.com
State New
Headers
Series iommu/dma: Document min_align_mask assumption |

Commit Message

Robin Murphy March 4, 2024, 12:05 p.m. UTC 
  iommu-dma does not explicitly reference min_align_mask since we already
assume that that will be less than or equal to any typical IOVA granule.
We wouldn't realistically expect to see the case where it is larger, and
that would be non-trivial to support, however for the sake of reasoning
(particularly around the interaction with SWIOTLB), let's clearly
enforce the assumption.

Signed-off-by: Robin Murphy <robin.murphy@arm.com>
---
 drivers/iommu/dma-iommu.c | 5 +++++
 1 file changed, 5 insertions(+)

Comments

Petr Tesařík March 4, 2024, 2:39 p.m. UTC | #1 
On Mon,  4 Mar 2024 12:05:42 +0000
Robin Murphy <robin.murphy@arm.com> wrote:

> iommu-dma does not explicitly reference min_align_mask since we already
> assume that that will be less than or equal to any typical IOVA granule.
> We wouldn't realistically expect to see the case where it is larger, and
> that would be non-trivial to support, however for the sake of reasoning
> (particularly around the interaction with SWIOTLB), let's clearly
> enforce the assumption.
> 
> Signed-off-by: Robin Murphy <robin.murphy@arm.com>

Looks good to me.

Reviewed-by: Petr Tesarik <petr.tesarik1@huawei-partners.com>

Thank you!

Petr T

> ---
>  drivers/iommu/dma-iommu.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
> index 50ccc4f1ef81..b58f5a3311c3 100644
> --- a/drivers/iommu/dma-iommu.c
> +++ b/drivers/iommu/dma-iommu.c
> @@ -859,6 +859,11 @@ static dma_addr_t __iommu_dma_map(struct device *dev, phys_addr_t phys,
>  	    iommu_deferred_attach(dev, domain))
>  		return DMA_MAPPING_ERROR;
>  
> +	/* If anyone ever wants this we'd need support in the IOVA allocator */
> +	if (dev_WARN_ONCE(dev, dma_get_min_align_mask(dev) > iova_mask(iovad),
> +	    "Unsupported alignment constraint\n"))
> +		return DMA_MAPPING_ERROR;
> +
>  	size = iova_align(iovad, size + iova_off);
>  
>  	iova = iommu_dma_alloc_iova(domain, size, dma_mask, dev);
Michael Kelley March 4, 2024, 3:47 p.m. UTC | #2 
From: Robin Murphy <robin.murphy@arm.com> Sent: Monday, March 4, 2024 4:06 AM
> 
> iommu-dma does not explicitly reference min_align_mask since we already
> assume that that will be less than or equal to any typical IOVA granule.
> We wouldn't realistically expect to see the case where it is larger, and
> that would be non-trivial to support, however for the sake of reasoning
> (particularly around the interaction with SWIOTLB), let's clearly
> enforce the assumption.
> 
> Signed-off-by: Robin Murphy <robin.murphy@arm.com>
> ---
>  drivers/iommu/dma-iommu.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
> index 50ccc4f1ef81..b58f5a3311c3 100644
> --- a/drivers/iommu/dma-iommu.c
> +++ b/drivers/iommu/dma-iommu.c
> @@ -859,6 +859,11 @@ static dma_addr_t __iommu_dma_map(struct device *dev, phys_addr_t phys,
>  	    iommu_deferred_attach(dev, domain))
>  		return DMA_MAPPING_ERROR;
> 
> +	/* If anyone ever wants this we'd need support in the IOVA allocator */
> +	if (dev_WARN_ONCE(dev, dma_get_min_align_mask(dev) > iova_mask(iovad),
> +	    "Unsupported alignment constraint\n"))
> +		return DMA_MAPPING_ERROR;
> +
>  	size = iova_align(iovad, size + iova_off);
> 
>  	iova = iommu_dma_alloc_iova(domain, size, dma_mask, dev);
> --
> 2.39.2.101.g768bb238c484.dirty

Reviewed-by: Michael Kelley <mhklinux@outlook.com>
diff mbox series

Patch

diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
index 50ccc4f1ef81..b58f5a3311c3 100644
--- a/drivers/iommu/dma-iommu.c
+++ b/drivers/iommu/dma-iommu.c
@@ -859,6 +859,11 @@  static dma_addr_t __iommu_dma_map(struct device *dev, phys_addr_t phys,
 	    iommu_deferred_attach(dev, domain))
 		return DMA_MAPPING_ERROR;
 
+	/* If anyone ever wants this we'd need support in the IOVA allocator */
+	if (dev_WARN_ONCE(dev, dma_get_min_align_mask(dev) > iova_mask(iovad),
+	    "Unsupported alignment constraint\n"))
+		return DMA_MAPPING_ERROR;
+
 	size = iova_align(iovad, size + iova_off);
 
 	iova = iommu_dma_alloc_iova(domain, size, dma_mask, dev);