On Wed, 24 Feb 2016 13:44:15 -0600
Bjorn Helgaas <bhelgaas@xxxxxxxxxx> wrote:
> This should be folded into the previous patch. I left it separate to show
> the interface difference more clearly.
>
> Also, pci_devs_are_dma_aliases() uses PCI internals (dma_alias_mask), so I
> think it should be in PCI code instead of in IOMMU code. That would mean
> both it and pci_add_dma_alias() should probably be declared in
> include/linux/pci.h (but not exported to modules with EXPORT_SYMBOL()).
> ---
> drivers/iommu/iommu.c | 11 ++++++-----
> 1 file changed, 6 insertions(+), 5 deletions(-)
Needs a Sign-off.
Reviewed-by: Alex Williamson <alex.williamson@xxxxxxxxxx>
>
> diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
> index a214e19..031d06b 100644
> --- a/drivers/iommu/iommu.c
> +++ b/drivers/iommu/iommu.c
> @@ -659,10 +659,12 @@ static struct iommu_group *get_pci_function_alias_group(struct pci_dev *pdev,
> return NULL;
> }
>
> -static bool dma_alias_is_enabled(struct pci_dev *dev, u8 devfn)
> +static bool pci_devs_are_dma_aliases(struct pci_dev *pdev1, struct pci_dev *pdev2)
> {
> - return dev->dma_alias_mask &&
> - test_bit(devfn, dev->dma_alias_mask);
> + return (pdev1->dma_alias_mask &&
> + test_bit(pdev2->devfn, pdev1->dma_alias_mask)) ||
> + (pdev2->dma_alias_mask &&
> + test_bit(pdev1->devfn, pdev2->dma_alias_mask));
> }
>
> /*
> @@ -692,8 +694,7 @@ static struct iommu_group *get_pci_alias_group(struct pci_dev *pdev,
> continue;
>
> /* We alias them or they alias us */
> - if (dma_alias_is_enabled(pdev, tmp->devfn) ||
> - dma_alias_is_enabled(tmp, pdev->devfn)) {
> + if (pci_devs_are_dma_aliases(pdev, tmp)) {
> group = get_pci_alias_group(tmp, devfns);
> if (group) {
> pci_dev_put(tmp);
>
--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
