RE: [PATCH v2 02/10] iommu: Introduce a new iommu_group_replace_domain() API

[Tian, Kevin]

> From: Jason Gunthorpe <jgg@xxxxxxxxxx>
> Sent: Thursday, February 9, 2023 9:23 PM
> 
> On Thu, Feb 09, 2023 at 02:55:24AM +0000, Tian, Kevin wrote:
> > > From: Nicolin Chen <nicolinc@xxxxxxxxxx>
> > > Sent: Wednesday, February 8, 2023 5:18 AM
> > >
> > > +int iommu_group_replace_domain(struct iommu_group *group,
> > > +			       struct iommu_domain *new_domain)
> > > +{
> > > +	int ret;
> > > +
> > > +	if (!new_domain)
> > > +		return -EINVAL;
> >
> > Is there value of allowing NULL new domain so this plays like
> > iommu_detach_group() then iommufd only needs call one
> > function in both attach/detach path?
> 
> We've used NULL to mean the 'platform domain' in the iommu core code
> in a few places, I'd prefer to avoid overloading NULL.
> 
> IMHO it doesn't help iommufd to substitute detach with replace.
> 

OK