Re: [PATCH 7/7] iommu/vtd: Split paging_domain_compatible()

[Baolu Lu <baolu.lu@linux.intel.com>]

On 6/11/25 07:51, Jason Gunthorpe wrote:
> On Tue, Jun 10, 2025 at 03:12:43PM +0800, Baolu Lu wrote:
>>> +	/*
>>> +	 * FIXME this is locked wrong, it needs to be under the
>>> +	 * dmar_domain->lock
>>> +	 */
>>>    	if (dmar_domain->force_snooping && !ecap_sc_support(iommu->ecap))
>>>    		return -EINVAL;
>>
>> Perhaps we can use group->mutex to fix this in the future?
> 
> It is part of the iommu_domain, there is no clear group->mutex that
> is affiliated with a domain, there can be multiple devices attached..

Fair enough.

> Maybe we should change the enforce coherency away from a function op
> and into a domain creation flag. Then it won't change at run time,
> fixing this race, and we don't have the issue of possibly having PTEs
> with the wrong bit already mapped into the domain.

That's a better approach. enforce_cache_coherency is only used in vfio
and iommufd, where the cache coherency is enforced just after the domain
is allocated.

> When this was first done that was harder, but now I think we have
> enough infrastructure to do it easily..

Yes. I can follow up with a patch series for broad discussion if you
don't want to include it in this series.

Thanks,
baolu