[PATCH rc] iommu/terga: Fix incorrect size calculation

[PATCH rc] iommu/terga: Fix incorrect size calculation

[Jason Gunthorpe]

This driver uses a mixture of ways to get the size of a PTE,
tegra_smmu_set_pde() did it as sizeof(*pd) which became wrong when pd
switched to a struct tegra_pd.

Switch pd back to a u32* in tegra_smmu_set_pde() so the sizeof(*pd)
returns 4.

Fixes: 50568f87d1e2 ("iommu/terga: Do not use struct page as the handle for as->pd memory")
Reported-by: Diogo Ivo <diogo.ivo@tecnico.ulisboa.pt>
Closes: https://lore.kernel.org/all/62e7f7fe-6200-4e4f-ad42-d58ad272baa6@tecnico.ulisboa.pt/
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/iommu/tegra-smmu.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c
index 61897d50162dd7..e58fe9d8b9e77e 100644
--- a/drivers/iommu/tegra-smmu.c
+++ b/drivers/iommu/tegra-smmu.c
@@ -559,11 +559,11 @@ static void tegra_smmu_set_pde(struct tegra_smmu_as *as, unsigned long iova,
 {
 	unsigned int pd_index = iova_pd_index(iova);
 	struct tegra_smmu *smmu = as->smmu;
-	struct tegra_pd *pd = as->pd;
+	u32 *pd = &as->pd->val[pd_index];
 	unsigned long offset = pd_index * sizeof(*pd);
 
 	/* Set the page directory entry first */
-	pd->val[pd_index] = value;
+	*pd = value;
 
 	/* The flush the page directory entry from caches */
 	dma_sync_single_range_for_device(smmu->dev, as->pd_dma, offset,

base-commit: 7d4dfa6140f75b67f8eff3ae6ebc9937be94c4d3
-- 
2.43.0

Re: [PATCH rc] iommu/terga: Fix incorrect size calculation

[Diogo Ivo]

On 6/3/25 8:14 PM, Jason Gunthorpe wrote:
> This driver uses a mixture of ways to get the size of a PTE,
> tegra_smmu_set_pde() did it as sizeof(*pd) which became wrong when pd
> switched to a struct tegra_pd.
> 
> Switch pd back to a u32* in tegra_smmu_set_pde() so the sizeof(*pd)
> returns 4.
> 
> Fixes: 50568f87d1e2 ("iommu/terga: Do not use struct page as the handle for as->pd memory")
> Reported-by: Diogo Ivo <diogo.ivo@tecnico.ulisboa.pt>
> Closes: https://lore.kernel.org/all/62e7f7fe-6200-4e4f-ad42-d58ad272baa6@tecnico.ulisboa.pt/
> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
> ---
>   drivers/iommu/tegra-smmu.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c
> index 61897d50162dd7..e58fe9d8b9e77e 100644
> --- a/drivers/iommu/tegra-smmu.c
> +++ b/drivers/iommu/tegra-smmu.c
> @@ -559,11 +559,11 @@ static void tegra_smmu_set_pde(struct tegra_smmu_as *as, unsigned long iova,
>   {
>   	unsigned int pd_index = iova_pd_index(iova);
>   	struct tegra_smmu *smmu = as->smmu;
> -	struct tegra_pd *pd = as->pd;
> +	u32 *pd = &as->pd->val[pd_index];
>   	unsigned long offset = pd_index * sizeof(*pd);
>   
>   	/* Set the page directory entry first */
> -	pd->val[pd_index] = value;
> +	*pd = value;
>   
>   	/* The flush the page directory entry from caches */
>   	dma_sync_single_range_for_device(smmu->dev, as->pd_dma, offset,
> 
> base-commit: 7d4dfa6140f75b67f8eff3ae6ebc9937be94c4d3

Just to be explicit about it:

Tested-by: Diogo Ivo <diogo.ivo@tecnico.ulisboa.pt>

Re: [PATCH rc] iommu/terga: Fix incorrect size calculation

[Jerry Snitselaar]

On Tue, Jun 03, 2025 at 04:14:45PM -0300, Jason Gunthorpe wrote:
> This driver uses a mixture of ways to get the size of a PTE,
> tegra_smmu_set_pde() did it as sizeof(*pd) which became wrong when pd
> switched to a struct tegra_pd.
> 
> Switch pd back to a u32* in tegra_smmu_set_pde() so the sizeof(*pd)
> returns 4.
> 
> Fixes: 50568f87d1e2 ("iommu/terga: Do not use struct page as the handle for as->pd memory")
> Reported-by: Diogo Ivo <diogo.ivo@tecnico.ulisboa.pt>
> Closes: https://lore.kernel.org/all/62e7f7fe-6200-4e4f-ad42-d58ad272baa6@tecnico.ulisboa.pt/
> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
> ---
>  drivers/iommu/tegra-smmu.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/iommu/tegra-smmu.c b/drivers/iommu/tegra-smmu.c
> index 61897d50162dd7..e58fe9d8b9e77e 100644
> --- a/drivers/iommu/tegra-smmu.c
> +++ b/drivers/iommu/tegra-smmu.c
> @@ -559,11 +559,11 @@ static void tegra_smmu_set_pde(struct tegra_smmu_as *as, unsigned long iova,
>  {
>  	unsigned int pd_index = iova_pd_index(iova);
>  	struct tegra_smmu *smmu = as->smmu;
> -	struct tegra_pd *pd = as->pd;
> +	u32 *pd = &as->pd->val[pd_index];
>  	unsigned long offset = pd_index * sizeof(*pd);
>  
>  	/* Set the page directory entry first */
> -	pd->val[pd_index] = value;
> +	*pd = value;
>  
>  	/* The flush the page directory entry from caches */
>  	dma_sync_single_range_for_device(smmu->dev, as->pd_dma, offset,
> 
> base-commit: 7d4dfa6140f75b67f8eff3ae6ebc9937be94c4d3
> -- 
> 2.43.0
> 

The tiniest of nits with the typo in the summary of terga vs tegra,
but other than that.

Reviewed-by: Jerry Snitselaar <jsnitsel@redhat.com>

Re: [PATCH rc] iommu/terga: Fix incorrect size calculation

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 777 bytes --]

On Tue, Jun 03, 2025 at 04:14:45PM -0300, Jason Gunthorpe wrote:
> This driver uses a mixture of ways to get the size of a PTE,
> tegra_smmu_set_pde() did it as sizeof(*pd) which became wrong when pd
> switched to a struct tegra_pd.
> 
> Switch pd back to a u32* in tegra_smmu_set_pde() so the sizeof(*pd)
> returns 4.
> 
> Fixes: 50568f87d1e2 ("iommu/terga: Do not use struct page as the handle for as->pd memory")
> Reported-by: Diogo Ivo <diogo.ivo@tecnico.ulisboa.pt>
> Closes: https://lore.kernel.org/all/62e7f7fe-6200-4e4f-ad42-d58ad272baa6@tecnico.ulisboa.pt/
> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
> ---
>  drivers/iommu/tegra-smmu.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

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

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

Re: [PATCH rc] iommu/terga: Fix incorrect size calculation

[Joerg Roedel]

On Tue, Jun 03, 2025 at 04:14:45PM -0300, Jason Gunthorpe wrote:
>  drivers/iommu/tegra-smmu.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

Applied, thanks.