RE: [PATCH v2 3/4] iommufd: Destroy vdevice on idevice destroy

[Tian, Kevin]

> From: Xu Yilun <yilun.xu@xxxxxxxxxxxxxxx>
> Sent: Tuesday, June 24, 2025 4:12 PM
> 
> On Tue, Jun 24, 2025 at 11:32:01AM +0800, Baolu Lu wrote:
> > On 6/23/25 17:49, Xu Yilun wrote:
> > > Destroy iommufd_vdevice(vdev) on iommufd_idevice(idev) destroy so
> that
> > > vdev can't outlive idev.
> > >
> > > iommufd_device(idev) represents the physical device bound to iommufd,
> > > while the iommufd_vdevice(vdev) represents the virtual instance of the
> > > physical device in the VM. The lifecycle of the vdev should not be
> > > longer than idev. This doesn't cause real problem on existing use cases
> > > cause vdev doesn't impact the physical device, only provides
> > > virtualization information. But to extend vdev for Confidential
> > > Computing(CC), there are needs to do secure configuration for the vdev,
> > > e.g. TSM Bind/Unbind. These configurations should be rolled back on idev
> > > destroy, or the external driver(VFIO) functionality may be impact.
> > >
> > > Building the association between idev & vdev requires the two objects
> > > pointing each other, but not referencing each other.
> >
> > Does this mean each idevice can have at most a single vdevice? Is it
> 
> Yes, I got this idea from here.
> 
> https://lore.kernel.org/all/20250604132403.GJ5028@xxxxxxxxxx/
> 
> > possible that different PASIDs of a physical device are assigned to
> > userspace for different purposes, such that there is a need for multiple
> > vdevices per idevice?
> 
> Mm.. I don't have a clear idea how SIOV assignment work with iommufd,
> may come back later.
> 

Let's put SIOV out of scope here. It's not supported yet. there are
other obstacles to be figured out (e.g. igroup etc.) when it comes to
the table.