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

[Jason Gunthorpe <jgg@nvidia.com>]

On Thu, Jun 26, 2025 at 11:31:06AM +0800, Xu Yilun wrote:
> > The wait_event is a ugly hack though, even in its existing code. The
> > above version is better because it doesn't have any failure mode and
> > doesn't introduce any unlocked use of the idev->vdev which is easier
> > to reason about, no READ_ONCE/WRITE_ONCE/etc
> > 
> > It sounds like you should largely leave the existing other parts the
> > same as this v2, though can you try reorganize it to look a little
> > more like the version I shared?
> 
> Sure. But may I confirm that your only want reentrant
> iommufd_vdevice_abort() but not your iommufd_object_remove_tombstone()
> changes?

I think take a look at how I organized the control flow in the patch I
sent and try to use some of those ideas, it was a bit simpler

> To me, grab a shortterm_users but not a user is a new operation model. I
> hesitate to add it when the existing refcount_inc(&obj->user) works for
> this case.

Yes, I am convinced you should not do this. Just hold the users only
and use the normal destroy with the XA_ZERO_ENTRY change

Along with the locked abort idea.

Jason