On Fri, Nov 01, 2024 at 12:18:50PM +0000, Will Deacon wrote: > On Wed, Oct 30, 2024 at 09:20:44PM -0300, Jason Gunthorpe wrote: > > [This is now based on Nicolin's iommufd patches for vIOMMU and will need > > to go through the iommufd tree, please ack] > > Can't we separate out the SMMUv3 driver changes? They shouldn't depend on > Nicolin's work afaict. We can put everything before "iommu/arm-smmu-v3: Support IOMMU_VIOMMU_ALLOC" directly on a rc and share a branch with your tree. That patch and following all depend on Nicolin's work, as they rely on the VIOMMU due to how different ARM is from Intel. How about you take these patches: [PATCH v4 01/12] vfio: Remove VFIO_TYPE1_NESTING_IOMMU [PATCH v4 02/12] ACPICA: IORT: Update for revision E.f [PATCH v4 03/12] ACPI/IORT: Support CANWBS memory access flag [PATCH v4 04/12] iommu/arm-smmu-v3: Report IOMMU_CAP_ENFORCE_CACHE_COHERENCY for CANWBS [PATCH v4 05/12] iommu/arm-smmu-v3: Support IOMMU_GET_HW_INFO via struct arm_smmu_hw_info [PATCH v4 06/12] iommu/arm-smmu-v3: Implement IOMMU_HWPT_ALLOC_NEST_PARENT [PATCH v4 07/12] iommu/arm-smmu-v3: Expose the arm_smmu_attach interface Onto a branch. I'll take these patches after merging your branch and Nicolin's: [PATCH v4 08/12] iommu/arm-smmu-v3: Support IOMMU_VIOMMU_ALLOC [PATCH v4 09/12] iommu/arm-smmu-v3: Support IOMMU_DOMAIN_NESTED [PATCH v4 10/12] iommu/arm-smmu-v3: Use S2FWB for NESTED domains [PATCH v4 11/12] iommu/arm-smmu-v3: Allow ATS for IOMMU_DOMAIN_NESTED [PATCH v4 12/12] iommu/arm-smmu-v3: Support IOMMU_HWPT_INVALIDATE using a VIOMMU object ? I can also probably push most of S2FWB and ATS into the first batch. Please let me know, I would like this to be done this cycle, Nicolin's vIOMMU series are all reviewed now. Jason