[PATCH 2/5] intel-iommu: Use dma_generic_alloc_coherent() for passthrough mappings

[Alex Williamson <alex.williamson@hp.com>]

intel_alloc_coherent() needs to follow DMA mapping convention and
make use of the coherent_dma_mask of the device.  Without this,
devices may get buffers they can't use.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
---

 drivers/pci/intel-iommu.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index b1e97e6..40be49a 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -2765,6 +2765,10 @@ static void *intel_alloc_coherent(struct device *hwdev, size_t size,
 	void *vaddr;
 	int order;
 
+	if (iommu_no_mapping(hwdev))
+		return dma_generic_alloc_coherent(hwdev, size, dma_handle,
+						  flags);
+
 	size = PAGE_ALIGN(size);
 	order = get_order(size);
 	flags &= ~(GFP_DMA | GFP_DMA32);
@@ -2788,6 +2792,10 @@ static void intel_free_coherent(struct device *hwdev, size_t size, void *vaddr,
 {
 	int order;
 
+	if (iommu_no_mapping(hwdev))
+		return dma_generic_free_coherent(hwdev, size, vaddr,
+						 dma_handle);
+
 	size = PAGE_ALIGN(size);
 	order = get_order(size);