Re: [PATCH v2 6/7] intel-iommu: Make use of DMA quirks and ACS checks in IOMMU groups

[Alex Williamson <alex.williamson-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>]

On Thu, 2012-05-31 at 15:47 -0400, Don Dutile wrote:
> On 05/30/2012 04:19 PM, Alex Williamson wrote:
> > Work around broken devices and adhere to ACS support when determining
> > IOMMU grouping.
> >
> > Signed-off-by: Alex Williamson<alex.williamson-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
> > ---
> >
> >   drivers/iommu/intel-iommu.c |   25 +++++++++++++++++++++++++
> >   1 file changed, 25 insertions(+)
> >
> > diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
> > index 4a43452..ebf2b31 100644
> > --- a/drivers/iommu/intel-iommu.c
> > +++ b/drivers/iommu/intel-iommu.c
> > @@ -4090,6 +4090,14 @@ static int intel_iommu_domain_has_cap(struct iommu_domain *domain,
> >   	return 0;
> >   }
> >
> > +static void swap_pci_ref(struct pci_dev **from, struct pci_dev *to)
> > +{
> > +	pci_dev_put(*from);
> > +	*from = to;
> > +}
> > +
> > +#define REQ_ACS_FLAGS	(PCI_ACS_SV | PCI_ACS_RR | PCI_ACS_CR | PCI_ACS_UF)
> > +
> >   static int intel_iommu_add_device(struct device *dev)
> >   {
> >   	struct pci_dev *pdev = to_pci_dev(dev);
> > @@ -4112,6 +4120,23 @@ static int intel_iommu_add_device(struct device *dev)
> >   	} else
> >   		dma_pdev = pci_dev_get(pdev);
> >
> > +	swap_pci_ref(&dma_pdev, pci_get_dma_source(dma_pdev));
> > +
> > +	if (dma_pdev->multifunction&&
> > +	    !pci_acs_enabled(dma_pdev, REQ_ACS_FLAGS))
> > +		swap_pci_ref(&dma_pdev,
> > +			     pci_get_slot(dma_pdev->bus,
> > +					  PCI_DEVFN(PCI_SLOT(dma_pdev->devfn),
> > +					  0)));
> > +
> > +	while (!pci_is_root_bus(dma_pdev->bus)) {
> > +		if (pci_acs_path_enabled(dma_pdev->bus->self,
> > +					 NULL, REQ_ACS_FLAGS))
> > +			break;
> > +
> > +		swap_pci_ref(&dma_pdev, pci_dev_get(dma_pdev->bus->self));
> > +	}
> > +
> I'm having deja-vu on this patch....
> .... why not just make the above two patches as two functions in drivers/iommu/iommu.c,
> one exported for these two modules (and maybe others someday...), e.g., iommu_pdev_put())
> which [intel-,amd-]iommu.c call ?

Do we want to put PCI specific code in the IOMMU base code?  I think
we'd want to avoid that.  I have no objection to reducing the
duplication, but I'm not sure who else is going to use this, where to
put it and what to call it.  So, I left the duplication here so it might
get more review from the IOMMU driver maintainers.  Thanks,

Alex