Re: [PATCH v3 1/5] iommu/amd: Remove iommu_v2 module

[Vasant Hegde <vasant.hegde@amd.com>]

Joerg,

On 9/25/2023 2:34 PM, joro@8bytes.org wrote:
> On Fri, Sep 22, 2023 at 09:18:12AM -0300, Jason Gunthorpe wrote:
>> IMHO iommu drivers should not override this.
>>
>> I think it was a mistake that iommu drivers could override to identity
>> domains. If we need quirks then the core code should have the quirk
>> list, similar to how PCI works.
> 
> Sounds reasonable, but if we transition over to an iommu core facility
> for quirks, we must do it in a way that does not cause temporary
> regressions for existing hardware.
> 
> That being said, Alexander, can you please test this series for
> regressions on the older hardware to make sure nothing regresses there?

This series just removes iommu_v2 module related code as we don't have any
in-kernel user of this module.

It doesn't modify how we allocate default domain for devices (GPUs with
PASID/PRI will be still in 1 x 1 mapping irrespective of boot iommu mode).
Hence IMO this series should not cause any regression.

I understand Alex's concern. We will work with Alex and make sure there is no
regression whenever we modify default domain allocation (1 x 1 mapping).

-Vasant