I can comment on one part here:
On Thu, Feb 08, 2018 at 12:18:44PM +0100, Christoffer Dall wrote:
> Wasn't this also the decision taken for IOMMU page table allocation, and
> why was that the right approach for the IOMMU but not for KVM stage 2
> page tables? Is there room for reuse of the IOMMU page table allocation
> logic in KVM as well?
There were a few reasons we did this for IOMMU page tables:
* Ability to use different page size/VA bits/levels from the CPU
* Ability to support different page table formats (e.g. short descriptor)
* Ability to determine page table attributes at runtime
* Requirement to map/unmap in atomic context
* Ability to cope with non-coherent page table walkers
* Ability to create both stage-1 and stage-2 mappings
* Easier to hook in our own TLB invalidation routines
* Support for lockless concurrent map/unmap within confines of the DMA API
usage
Will
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
