RE: [PATCH v10 6/7] iommu/vt-d: Add set_dev_pasid callback for dma domain

[Tian, Kevin]

> From: Baolu Lu <baolu.lu@xxxxxxxxxxxxxxx>
> Sent: Friday, July 14, 2023 2:00 PM
> 
> On 2023/7/14 11:50, Tian, Kevin wrote:
> >> From: Baolu Lu <baolu.lu@xxxxxxxxxxxxxxx>
> >> Sent: Friday, July 14, 2023 11:34 AM
> >>
> >> On 2023/7/13 15:56, Tian, Kevin wrote:
> >>>> From: Jacob Pan <jacob.jun.pan@xxxxxxxxxxxxxxx>
> >>>> Sent: Thursday, July 13, 2023 12:34 AM
> >>>>
> >>>>
> >>>> -	/*
> >>>> -	 * Should never reach here until we add support for attaching
> >>>> -	 * non-SVA domain to a pasid.
> >>>> -	 */
> >>>> -	WARN_ON(1);
> >>>> +	dmar_domain = to_dmar_domain(domain);
> >>>> +	spin_lock_irqsave(&dmar_domain->lock, flags);
> >>>> +	list_for_each_entry(curr, &dmar_domain->dev_pasids, link_domain)
> >>>> {
> >>>> +		if (curr->dev == dev && curr->pasid == pasid) {
> >>>> +			list_del(&curr->link_domain);
> >>>> +			dev_pasid = curr;
> >>>> +			break;
> >>>> +		}
> >>>> +	}
> >>>> +	spin_unlock_irqrestore(&dmar_domain->lock, flags);
> >>>>
> >>>
> >>> what about no matching dev_pasid is find?
> >>
> >> kfree() can handle this gracefully.
> >
> > but what about domain_detach_iommu()? Is it correct to adjust
> > the refcnting when there is no matching dev_pasid?
> 
> You are right.
> 
> Logically, if we get a valid domain for a pasid, we should have a
> dev_pasid allocated for it. Perhaps, adding a check in the code will
> make the code more readable?
> 

yes