Re: [PATCH v1 3/3] iommu/arm-smmu-v3-sva: Remove arm_smmu_bond

[Jason Gunthorpe]

On Tue, Sep 05, 2023 at 09:14:09PM +0800, Michael Shavit wrote:
> On Tue, Sep 5, 2023 at 8:42 PM Jason Gunthorpe <jgg@xxxxxxxxxx> wrote:
> >
> > On Tue, Sep 05, 2023 at 07:49:14PM +0800, Michael Shavit wrote:
> > > Create a new iommu_domain subclass for SVA iommu domains to hold the
> > > data previously stored in the dynamically allocated arm_smmu_bond. Add a
> > > simple count of attached SVA domains to arm_smmu_master to replace the
> > > list of bonds.
> > >
> > > Signed-off-by: Michael Shavit <mshavit@xxxxxxxxxx>
> > > ---
> > >
> > >  .../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c   | 70 +++++++------------
> > >  drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c   |  1 -
> > >  drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h   |  2 +-
> > >  3 files changed, 26 insertions(+), 47 deletions(-)
> > >
> > > diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c
> > > index 9fb6907c5e7d4..0342c0f35d55a 100644
> > > --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c
> > > +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c
> > > @@ -24,14 +24,13 @@ struct arm_smmu_mmu_notifier {
> > >
> > >  #define mn_to_smmu(mn) container_of(mn, struct arm_smmu_mmu_notifier, mn)
> > >
> > > -struct arm_smmu_bond {
> > > -     struct mm_struct                *mm;
> > > +struct arm_smmu_sva_domain {
> > > +     struct iommu_domain             iommu_domain;
> > >       struct arm_smmu_mmu_notifier    *smmu_mn;
> > > -     struct list_head                list;
> > >  };
> > >
> > > -#define sva_to_bond(handle) \
> > > -     container_of(handle, struct arm_smmu_bond, sva)
> > > +#define to_sva_domain(domain) \
> > > +     container_of(domain, struct arm_smmu_sva_domain, iommu_domain)
> >
> > I'm not sure about this? This seems like a strange direction
> >
> > The SVA domain and a UNMANAGED/PAGING domain should be basically the
> > same thing. Making a sva_domain a completely different type looks like
> > it would stand in the way of that?
> 
> Agreed that's the eventual destination of all these re-works, but the
> stage isn't fully set for that yet. IMO this is a simpler improvement
> to get through for now, and I don't see it being an obstacle in the
> future.

Well, OK, you have the followup patches..

But I don't want to get in a spot where we continue to have "primary
domains" for SVA..

Jason