Re: [PATCH v2 02/10] iommu: Introduce a new iommu_group_replace_domain() API

[Jason Gunthorpe <jgg@nvidia.com>]

On Wed, Feb 22, 2023 at 02:11:39AM +0000, Tian, Kevin wrote:

> > > There was an attempt [1] to fix error unwind in iommu_attach_group(), by
> > > temporarily set group->domain to NULL before calling set_domain().
> > >
> > > Jason, I wonder why this recovering cannot be done in
> > > __iommu_group_set_domain() directly, e.g.:
> > >
> > > 	ret = __iommu_group_for_each_dev(group, new_domain,
> > > 					  iommu_group_do_attach_device);
> > > 	if (ret) {
> > > 		__iommu_group_for_each_dev(group, group->domain,
> > > 					  iommu_group_do_attach_device);
> > > 		return ret;
> > > 	}
> > > 	group->domain = new_domain;
> > 
> > We talked about this already, some times this is not the correct
> > recovery case, eg if we are going to a blocking domain we need to drop
> > all references to the prior domain, not put them back.
> > 
> > Failures are WARN_ON events not error recovery.
> > 
> 
> OK, I remember that. Then here looks we also need temporarily
> set group->domain to NULL before calling set_domain() to recover,
> as [1] does.

Sigh, this is too much.

I made a series to clean up all the domain attach logic so the error
handling is all in one place and all the same.

What do you think?

https://github.com/jgunthorpe/linux/commits/iommufd_hwpt

Jason