Re: [RFC PATCH v2 12/15] vfio: Support VFIO devices that access the CPU MMU

[Alex Williamson <alex@shazbot.org>]

On Tue, 20 Oct 2026 16:06:37 +0100
Ryan Roberts <ryan.roberts@arm.com> wrote:

> On 20/10/2026 15:09, Jason Gunthorpe wrote:
> > On Tue, Oct 20, 2026 at 12:51:26PM +0100, Ryan Roberts wrote:  
> >>> However.. you still have to affirm which single mm is the mm that the
> >>> device is going to be doing DMA against. The process the opens the fd
> >>> should not be forced to be the mm that it gets linked to.  
> >>
> >> What is the mm representing here? Ultimately we want to install the KVM
> >> instance's stage2 pgtables and (pinned) VMID. We currently get that via the
> >> bound KVM instance (which looks like it still works using the cdev interface?)
> >>
> >> The guest will then install the stage1 pgtables and (pinned) ASID associated
> >> with the relevant (guest) mm_struct.
> >>
> >> Do we _need_ an mm association in the host?  
> > 
> > Ah.. I see, I didn't quite follow that. OK.. Ah everyone else doing
> > this is using iommufd with a viommu to establish their connection
> > between device an KVM. eg Jacob Pan's series. That's a probably a
> > better idea to stay aligned with the common path. Though your reasons
> > are a little different..  
> 
> Just to check, you're referring to [1]?
> 
> I was under the impression that I could open the VFIO device through the cdev
> interface, then bind it to the KVM instance via KVM_DEV_VFIO_FILE_ADD... Why do
> I need an iommufd at all?

The cdev is not unlocked to actually access the device until it's bound
to iommufd.  Thanks,

Alex