[PATCH v1 08/12] iommufd/viommu: Replace ops->viommu_alloc with ops->viommu_init

[Tian, Kevin]

> From: Nicolin Chen <nicolinc at nvidia.com>
> Sent: Tuesday, June 10, 2025 2:20 PM
> 
> On Tue, Jun 10, 2025 at 01:55:05PM +0800, Baolu Lu wrote:
> > On 6/10/25 01:13, Nicolin Chen wrote:
> > > To ease the for-driver iommufd APIs, get_viommu_size and viommu_init
> ops
> > > are introduced. Now, those existing vIOMMU supported drivers
> implemented
> > > these two ops too.
> > >
> > > Relace the ops->viommu_alloc call with the two new ones.
> > >
> > > Note that this will fail a !viommu->ops case from now on, since a
> vIOMMU
> > > is expected to support alloc_domain_nested at least.
> >
> > Does this mean that the viommu implementation in the iommu driver is
> > required to implement alloc_domain_nested? I suppose viommu should
> soon
> > be extended to support TEE/IO.
> 
> It's a good point that CCA might not need a nested domain. So,
> it's inaccurate to say that, although I suspect that CCA would
> need some other viommu op then the check here would be sane.
> 
> With that being said, it's probably not worth adding that until
> we are 100% sure that no case will work with a !viommu->ops, so
> let's drop this new rejection, since we haven't had it so far.
> 

WARN_ON_ONCE() is built on the current context, not the future
usage. So I'd prefer to keeping it until there is a real need to revert.

Reviewed-by: Kevin Tian <kevin.tian at intel.com>