drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
PAGE_SIZE can be 16KB for Tegra which is not supported by MMU-500 on
both Tegra194 and Tegra234. So, use 4KB mappings when PAGE_SIZE is 16KB.
Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
---
drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
index 4b2994b6126d..bb621a94f6fe 100644
--- a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
+++ b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
@@ -273,11 +273,12 @@ static int nvidia_smmu_init_context(struct arm_smmu_domain *smmu_domain,
* release of PMD entry and avoid translations seeing stale PMD entry in
* walk cache.
* Fix this by limiting the page mappings to PAGE_SIZE on Tegra194 and
- * Tegra234.
+ * Tegra234. Use 4K page mappings if PAGE_SIZE is 16K as MMU500 doesn't
+ * support it.
*/
if (of_device_is_compatible(np, "nvidia,tegra234-smmu") ||
of_device_is_compatible(np, "nvidia,tegra194-smmu")) {
- smmu->pgsize_bitmap = PAGE_SIZE;
+ smmu->pgsize_bitmap = (PAGE_SIZE == SZ_16K) ? SZ_4K : PAGE_SIZE;
pgtbl_cfg->pgsize_bitmap = smmu->pgsize_bitmap;
}
--
2.25.1
On Wed, Jul 17, 2024 at 07:21:45AM +0000, Ashish Mhetre wrote:
> PAGE_SIZE can be 16KB for Tegra which is not supported by MMU-500 on
> both Tegra194 and Tegra234. So, use 4KB mappings when PAGE_SIZE is 16KB.
>
> Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
> ---
> drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
> index 4b2994b6126d..bb621a94f6fe 100644
> --- a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
> +++ b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
> @@ -273,11 +273,12 @@ static int nvidia_smmu_init_context(struct arm_smmu_domain *smmu_domain,
> * release of PMD entry and avoid translations seeing stale PMD entry in
> * walk cache.
> * Fix this by limiting the page mappings to PAGE_SIZE on Tegra194 and
> - * Tegra234.
> + * Tegra234. Use 4K page mappings if PAGE_SIZE is 16K as MMU500 doesn't
> + * support it.
> */
> if (of_device_is_compatible(np, "nvidia,tegra234-smmu") ||
> of_device_is_compatible(np, "nvidia,tegra194-smmu")) {
> - smmu->pgsize_bitmap = PAGE_SIZE;
> + smmu->pgsize_bitmap = (PAGE_SIZE == SZ_16K) ? SZ_4K : PAGE_SIZE;
Perhaps it would be more robust to do a read-modify-write of the
pgsize_bitmap, preserving all the page-sizes that are <= PAGE_SIZE (i.e.
and it with something like GENMASK(PAGE_SHIFT, 0)?
Will
On 7/23/2024 9:41 PM, Will Deacon wrote:
> External email: Use caution opening links or attachments
>
>
> On Wed, Jul 17, 2024 at 07:21:45AM +0000, Ashish Mhetre wrote:
>> PAGE_SIZE can be 16KB for Tegra which is not supported by MMU-500 on
>> both Tegra194 and Tegra234. So, use 4KB mappings when PAGE_SIZE is 16KB.
>>
>> Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
>> ---
>> drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c | 5 +++--
>> 1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
>> index 4b2994b6126d..bb621a94f6fe 100644
>> --- a/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
>> +++ b/drivers/iommu/arm/arm-smmu/arm-smmu-nvidia.c
>> @@ -273,11 +273,12 @@ static int nvidia_smmu_init_context(struct arm_smmu_domain *smmu_domain,
>> * release of PMD entry and avoid translations seeing stale PMD entry in
>> * walk cache.
>> * Fix this by limiting the page mappings to PAGE_SIZE on Tegra194 and
>> - * Tegra234.
>> + * Tegra234. Use 4K page mappings if PAGE_SIZE is 16K as MMU500 doesn't
>> + * support it.
>> */
>> if (of_device_is_compatible(np, "nvidia,tegra234-smmu") ||
>> of_device_is_compatible(np, "nvidia,tegra194-smmu")) {
>> - smmu->pgsize_bitmap = PAGE_SIZE;
>> + smmu->pgsize_bitmap = (PAGE_SIZE == SZ_16K) ? SZ_4K : PAGE_SIZE;
> Perhaps it would be more robust to do a read-modify-write of the
> pgsize_bitmap, preserving all the page-sizes that are <= PAGE_SIZE (i.e.
> and it with something like GENMASK(PAGE_SHIFT, 0)?
>
> Will
Thanks for the review Will. Yes, that will be a better approach as
pgsize_bitmap will never be with 16K granularity for Tegra.
I will send new version with this update.
© 2016 - 2025 Red Hat, Inc.