On Wed, Jul 26, 2023 at 08:36:31PM -0300, Jason Gunthorpe wrote:
> On Wed, Jul 26, 2023 at 01:50:28PM -0700, Nicolin Chen wrote:
> >
> > > > rc = iopt_add_access(&new_ioas->iopt, access);
> > > > if (rc) {
> > > > - mutex_unlock(&access->ioas_lock);
> > > > iommufd_put_object(&new_ioas->obj);
> > > > + if (cur_ioas)
> > > > + WARN_ON(iommufd_access_change_pt(access,
> > > > + cur_ioas->obj.id));
> > >
> > > We've already dropped our ref to cur_ioas, so this is also racy with
> > > destroy.
> >
> > Would it be better by calling iommufd_access_detach() that holds
> > the same mutex in the iommufd_access_destroy_object()? We could
> > also unwrap the detach and delay the refcount_dec, as you did in
> > your attaching patch.
>
> It is better just to integrate it with this algorithm so we don't have
> the refcounting issues, like I did
OK. I will have a patch adding the iommufd_access_change_ioas
first, and it can update iommufd_access_destroy_object() too.
> > > This is what I came up with:
> > >
> > > diff --git a/drivers/iommu/iommufd/device.c b/drivers/iommu/iommufd/device.c
> > > index 57c0e81f5073b2..e55d6e902edb98 100644
> > > --- a/drivers/iommu/iommufd/device.c
> > > +++ b/drivers/iommu/iommufd/device.c
> > > @@ -758,64 +758,101 @@ void iommufd_access_destroy(struct iommufd_access *access)
> > > }
> > > EXPORT_SYMBOL_NS_GPL(iommufd_access_destroy, IOMMUFD);
> > >
> > > -void iommufd_access_detach(struct iommufd_access *access)
> > > +static int iommufd_access_change_ioas(struct iommufd_access *access,
> > > + struct iommufd_ioas *new_ioas)
> > > {
> > > struct iommufd_ioas *cur_ioas = access->ioas;
> > > + int rc;
> > > +
> > > + lockdep_assert_held(&access->ioas_lock);
> > > +
> > > + /* We are racing with a concurrent detach, bail */
> > > + if (access->ioas_unpin)
> > > + return -EBUSY;
> >
> > I think this should check access->ioas too? I mean:
>
> >
> > + /* We are racing with a concurrent detach, bail */
> > + if (!access->ioas && access->ioas_unpin)
> > + return -EBUSY;
>
> Oh, yes, that should basically be 'cur_ioas != access->ioas_unpin' -
> ie any difference means we are racing with the unmap call.
Yea, will update to 'cur_ioas != access->ioas_unpin'.
> > > + if (new_ioas) {
> > > + rc = iopt_add_access(&new_ioas->iopt, access);
> > > + if (rc) {
> > > + iommufd_put_object(&new_ioas->obj);
> > > + access->ioas = cur_ioas;
> > > + return rc;
> > > + }
> > > + iommufd_ref_to_users(&new_ioas->obj);
> > > + }
> > > +
> > > + access->ioas = new_ioas;
> > > + access->ioas_unpin = new_ioas;
> > > iopt_remove_access(&cur_ioas->iopt, access);
> >
> > There was a bug in my earlier version, having the same flow by
> > calling iopt_add_access() prior to iopt_remove_access(). But,
> > doing that would override the access->iopt_access_list_id and
> > it would then get unset by the following iopt_remove_access().
>
> Ah, I was wondering about that order but didn't check it.
>
> Maybe we just need to pass the ID into iopt_remove_access and keep the
> right version on the stack.
>
> > So, I came up with this version calling an iopt_remove_access()
> > prior to iopt_add_access(), which requires an add-back the old
> > ioas upon an failure at iopt_add_access(new_ioas).
>
> That is also sort of reasonable if the refcounting is organized like
> this does.
I just realized that either my v8 or your version calls unmap()
first at the entire cur_ioas. So, there seems to be no point in
doing that fallback re-add routine since the cur_ioas isn't the
same, which I don't feel quite right...
Perhaps we should pass the ID into iopt_add/remove_access like
you said above. And then we attach the new_ioas, piror to the
detach the cur_ioas?
Thanks
Nicolin
