Re: [PATCH v2 10/12] iommu: Make iommu_queue_iopf() more generic

[Baolu Lu]

On 2023/8/11 3:07, Jason Gunthorpe wrote:
> On Thu, Jul 27, 2023 at 01:48:35PM +0800, Lu Baolu wrote:
> > @@ -137,6 +136,16 @@ int iommu_queue_iopf(struct iommu_fault *fault, struct device *dev)
> >   		return 0;
> >   	}
> >   
> > +	if (fault->prm.flags & IOMMU_FAULT_PAGE_REQUEST_PASID_VALID)
> > +		domain = iommu_get_domain_for_dev_pasid(dev, fault->prm.pasid, 0);
> > +	else
> > +		domain = iommu_get_domain_for_dev(dev);
>
> How does the lifetime work for this? What prevents UAF on domain?

Replied below.

> > diff --git a/drivers/iommu/iommu-sva.c b/drivers/iommu/iommu-sva.c
> > index ab42cfdd7636..668f4c2bcf65 100644
> > --- a/drivers/iommu/iommu-sva.c
> > +++ b/drivers/iommu/iommu-sva.c
> > @@ -157,7 +157,7 @@ EXPORT_SYMBOL_GPL(iommu_sva_get_pasid);
> >   /*
> >    * I/O page fault handler for SVA
> >    */
> > -enum iommu_page_response_code
> > +static enum iommu_page_response_code
> >   iommu_sva_handle_iopf(struct iommu_fault *fault, void *data)
> >   {
> >   	vm_fault_t ret;
> > @@ -241,23 +241,16 @@ static void iopf_handler(struct work_struct *work)
> >   {
> >   	struct iopf_fault *iopf;
> >   	struct iopf_group *group;
> > -	struct iommu_domain *domain;
> >   	enum iommu_page_response_code status = IOMMU_PAGE_RESP_SUCCESS;
> >   
> >   	group = container_of(work, struct iopf_group, work);
> > -	domain = iommu_get_domain_for_dev_pasid(group->dev,
> > -				group->last_fault.fault.prm.pasid, 0);
> > -	if (!domain || !domain->iopf_handler)
> > -		status = IOMMU_PAGE_RESP_INVALID;
> > -
> >   	list_for_each_entry(iopf, &group->faults, list) {
> >   		/*
> >   		 * For the moment, errors are sticky: don't handle subsequent
> >   		 * faults in the group if there is an error.
> >   		 */
> >   		if (status == IOMMU_PAGE_RESP_SUCCESS)
> > -			status = domain->iopf_handler(&iopf->fault,
> > -						      domain->fault_data);
> > +			status = iommu_sva_handle_iopf(&iopf->fault, group->data);
> >   	}
> >   
> >   	iopf_complete_group(group->dev, &group->last_fault, status);
> > diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
> > index 157a28a49473..535a36e3edc9 100644
> > --- a/drivers/iommu/iommu.c
> > +++ b/drivers/iommu/iommu.c
> > @@ -3330,7 +3330,7 @@ struct iommu_domain *iommu_sva_domain_alloc(struct device *dev,
> >   	domain->type = IOMMU_DOMAIN_SVA;
> >   	mmgrab(mm);
> >   	domain->mm = mm;
> > -	domain->iopf_handler = iommu_sva_handle_iopf;
> > +	domain->iopf_handler = iommu_sva_handle_iopf_group;
> >   	domain->fault_data = mm;
>
> This also has lifetime problems on the mm.
>
> The domain should flow into the iommu_sva_handle_iopf() instead of the
> void *data.

Okay, but I still want to keep void *data as a private pointer of the
iopf consumer. For SVA, it's probably NULL.

> The SVA code can then just use domain->mm directly.

Yes.

> We need to document/figure out some how to ensure that the faults are
> all done processing before a fault enabled domain can be freed.

This has been documented in drivers/iommu/io-pgfault.c:

[...]
 * Any valid page fault will be eventually routed to an iommu domain 
and the
 * page fault handler installed there will get called. The users of this
 * handling framework should guarantee that the iommu domain could only be
 * freed after the device has stopped generating page faults (or the iommu
 * hardware has been set to block the page faults) and the pending page 
faults
 * have been flushed.
 *
 * Return: 0 on success and <0 on error.
 */
int iommu_queue_iopf(struct iommu_fault *fault, void *cookie)
[...]

> This patch would be better ordered before the prior patch.

Let me try this in the next version.

Best regards,
baolu