Merge branch 'iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (60 commits)
  dma-debug: make memory range checks more consistent
  dma-debug: warn of unmapping an invalid dma address
  dma-debug: fix dma_debug_add_bus() definition for !CONFIG_DMA_API_DEBUG
  dma-debug/x86: register pci bus for dma-debug leak detection
  dma-debug: add a check dma memory leaks
  dma-debug: add checks for kernel text and rodata
  dma-debug: print stacktrace of mapping path on unmap error
  dma-debug: Documentation update
  dma-debug: x86 architecture bindings
  dma-debug: add function to dump dma mappings
  dma-debug: add checks for sync_single_sg_*
  dma-debug: add checks for sync_single_range_*
  dma-debug: add checks for sync_single_*
  dma-debug: add checking for [alloc|free]_coherent
  dma-debug: add add checking for map/unmap_sg
  dma-debug: add checking for map/unmap_page/single
  dma-debug: add core checking functions
  dma-debug: add debugfs interface
  dma-debug: add kernel command line parameters
  dma-debug: add initialization code
  ...

Fix trivial conflicts due to whitespace changes in arch/x86/kernel/pci-nommu.c

28 files changed, 502 insertions, 747 deletions

diff --git a/arch/Kconfig b/arch/Kconfig
index 550dab22daa1..830c16a2b801 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -106,3 +106,5 @@ config HAVE_CLK
 	  The <linux/clk.h> calls support software clock gating and
 	  thus are a key power management tool on many systems.
 
+config HAVE_DMA_API_DEBUG
+	bool
diff --git a/arch/ia64/dig/Makefile b/arch/ia64/dig/Makefile
index 5c0283830bd6..2f7caddf093e 100644
--- a/arch/ia64/dig/Makefile
+++ b/arch/ia64/dig/Makefile
@@ -7,8 +7,8 @@
 
 obj-y := setup.o
 ifeq ($(CONFIG_DMAR), y)
-obj-$(CONFIG_IA64_GENERIC) += machvec.o machvec_vtd.o dig_vtd_iommu.o
+obj-$(CONFIG_IA64_GENERIC) += machvec.o machvec_vtd.o
 else
 obj-$(CONFIG_IA64_GENERIC) += machvec.o
 endif
-obj-$(CONFIG_IA64_DIG_VTD) += dig_vtd_iommu.o
+
diff --git a/arch/ia64/dig/dig_vtd_iommu.c b/arch/ia64/dig/dig_vtd_iommu.c
deleted file mode 100644
index 1c8a079017a3..000000000000
--- a/arch/ia64/dig/dig_vtd_iommu.c
+++ /dev/null
@@ -1,59 +0,0 @@
-#include <linux/types.h>
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/intel-iommu.h>
-
-void *
-vtd_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
-		 gfp_t flags)
-{
-	return intel_alloc_coherent(dev, size, dma_handle, flags);
-}
-EXPORT_SYMBOL_GPL(vtd_alloc_coherent);
-
-void
-vtd_free_coherent(struct device *dev, size_t size, void *vaddr,
-		 dma_addr_t dma_handle)
-{
-	intel_free_coherent(dev, size, vaddr, dma_handle);
-}
-EXPORT_SYMBOL_GPL(vtd_free_coherent);
-
-dma_addr_t
-vtd_map_single_attrs(struct device *dev, void *addr, size_t size,
-		     int dir, struct dma_attrs *attrs)
-{
-	return intel_map_single(dev, (phys_addr_t)addr, size, dir);
-}
-EXPORT_SYMBOL_GPL(vtd_map_single_attrs);
-
-void
-vtd_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
-		       int dir, struct dma_attrs *attrs)
-{
-	intel_unmap_single(dev, iova, size, dir);
-}
-EXPORT_SYMBOL_GPL(vtd_unmap_single_attrs);
-
-int
-vtd_map_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
-		 int dir, struct dma_attrs *attrs)
-{
-	return intel_map_sg(dev, sglist, nents, dir);
-}
-EXPORT_SYMBOL_GPL(vtd_map_sg_attrs);
-
-void
-vtd_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
-		   int nents, int dir, struct dma_attrs *attrs)
-{
-	intel_unmap_sg(dev, sglist, nents, dir);
-}
-EXPORT_SYMBOL_GPL(vtd_unmap_sg_attrs);
-
-int
-vtd_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
-{
-	return 0;
-}
-EXPORT_SYMBOL_GPL(vtd_dma_mapping_error);
diff --git a/arch/ia64/hp/common/hwsw_iommu.c b/arch/ia64/hp/common/hwsw_iommu.c
index 2769dbfd03bf..e4a80d82e3d8 100644
--- a/arch/ia64/hp/common/hwsw_iommu.c
+++ b/arch/ia64/hp/common/hwsw_iommu.c
@@ -13,49 +13,34 @@
  */
 
 #include <linux/device.h>
+#include <linux/dma-mapping.h>
 #include <linux/swiotlb.h>
-
 #include <asm/machvec.h>
 
+extern struct dma_map_ops sba_dma_ops, swiotlb_dma_ops;
+
 /* swiotlb declarations & definitions: */
 extern int swiotlb_late_init_with_default_size (size_t size);
 
-/* hwiommu declarations & definitions: */
-
-extern ia64_mv_dma_alloc_coherent	sba_alloc_coherent;
-extern ia64_mv_dma_free_coherent	sba_free_coherent;
-extern ia64_mv_dma_map_single_attrs	sba_map_single_attrs;
-extern ia64_mv_dma_unmap_single_attrs	sba_unmap_single_attrs;
-extern ia64_mv_dma_map_sg_attrs		sba_map_sg_attrs;
-extern ia64_mv_dma_unmap_sg_attrs	sba_unmap_sg_attrs;
-extern ia64_mv_dma_supported		sba_dma_supported;
-extern ia64_mv_dma_mapping_error	sba_dma_mapping_error;
-
-#define hwiommu_alloc_coherent		sba_alloc_coherent
-#define hwiommu_free_coherent		sba_free_coherent
-#define hwiommu_map_single_attrs	sba_map_single_attrs
-#define hwiommu_unmap_single_attrs	sba_unmap_single_attrs
-#define hwiommu_map_sg_attrs		sba_map_sg_attrs
-#define hwiommu_unmap_sg_attrs		sba_unmap_sg_attrs
-#define hwiommu_dma_supported		sba_dma_supported
-#define hwiommu_dma_mapping_error	sba_dma_mapping_error
-#define hwiommu_sync_single_for_cpu	machvec_dma_sync_single
-#define hwiommu_sync_sg_for_cpu		machvec_dma_sync_sg
-#define hwiommu_sync_single_for_device	machvec_dma_sync_single
-#define hwiommu_sync_sg_for_device	machvec_dma_sync_sg
-
-
 /*
  * Note: we need to make the determination of whether or not to use
  * the sw I/O TLB based purely on the device structure.  Anything else
  * would be unreliable or would be too intrusive.
  */
-static inline int
-use_swiotlb (struct device *dev)
+static inline int use_swiotlb(struct device *dev)
 {
-	return dev && dev->dma_mask && !hwiommu_dma_supported(dev, *dev->dma_mask);
+	return dev && dev->dma_mask &&
+		!sba_dma_ops.dma_supported(dev, *dev->dma_mask);
 }
 
+struct dma_map_ops *hwsw_dma_get_ops(struct device *dev)
+{
+	if (use_swiotlb(dev))
+		return &swiotlb_dma_ops;
+	return &sba_dma_ops;
+}
+EXPORT_SYMBOL(hwsw_dma_get_ops);
+
 void __init
 hwsw_init (void)
 {
@@ -71,125 +56,3 @@ hwsw_init (void)
 #endif
 	}
 }
-
-void *
-hwsw_alloc_coherent (struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t flags)
-{
-	if (use_swiotlb(dev))
-		return swiotlb_alloc_coherent(dev, size, dma_handle, flags);
-	else
-		return hwiommu_alloc_coherent(dev, size, dma_handle, flags);
-}
-
-void
-hwsw_free_coherent (struct device *dev, size_t size, void *vaddr, dma_addr_t dma_handle)
-{
-	if (use_swiotlb(dev))
-		swiotlb_free_coherent(dev, size, vaddr, dma_handle);
-	else
-		hwiommu_free_coherent(dev, size, vaddr, dma_handle);
-}
-
-dma_addr_t
-hwsw_map_single_attrs(struct device *dev, void *addr, size_t size, int dir,
-		       struct dma_attrs *attrs)
-{
-	if (use_swiotlb(dev))
-		return swiotlb_map_single_attrs(dev, addr, size, dir, attrs);
-	else
-		return hwiommu_map_single_attrs(dev, addr, size, dir, attrs);
-}
-EXPORT_SYMBOL(hwsw_map_single_attrs);
-
-void
-hwsw_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
-			 int dir, struct dma_attrs *attrs)
-{
-	if (use_swiotlb(dev))
-		return swiotlb_unmap_single_attrs(dev, iova, size, dir, attrs);
-	else
-		return hwiommu_unmap_single_attrs(dev, iova, size, dir, attrs);
-}
-EXPORT_SYMBOL(hwsw_unmap_single_attrs);
-
-int
-hwsw_map_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
-		   int dir, struct dma_attrs *attrs)
-{
-	if (use_swiotlb(dev))
-		return swiotlb_map_sg_attrs(dev, sglist, nents, dir, attrs);
-	else
-		return hwiommu_map_sg_attrs(dev, sglist, nents, dir, attrs);
-}
-EXPORT_SYMBOL(hwsw_map_sg_attrs);
-
-void
-hwsw_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
-		     int dir, struct dma_attrs *attrs)
-{
-	if (use_swiotlb(dev))
-		return swiotlb_unmap_sg_attrs(dev, sglist, nents, dir, attrs);
-	else
-		return hwiommu_unmap_sg_attrs(dev, sglist, nents, dir, attrs);
-}
-EXPORT_SYMBOL(hwsw_unmap_sg_attrs);
-
-void
-hwsw_sync_single_for_cpu (struct device *dev, dma_addr_t addr, size_t size, int dir)
-{
-	if (use_swiotlb(dev))
-		swiotlb_sync_single_for_cpu(dev, addr, size, dir);
-	else
-		hwiommu_sync_single_for_cpu(dev, addr, size, dir);
-}
-
-void
-hwsw_sync_sg_for_cpu (struct device *dev, struct scatterlist *sg, int nelems, int dir)
-{
-	if (use_swiotlb(dev))
-		swiotlb_sync_sg_for_cpu(dev, sg, nelems, dir);
-	else
-		hwiommu_sync_sg_for_cpu(dev, sg, nelems, dir);
-}
-
-void
-hwsw_sync_single_for_device (struct device *dev, dma_addr_t addr, size_t size, int dir)
-{
-	if (use_swiotlb(dev))
-		swiotlb_sync_single_for_device(dev, addr, size, dir);
-	else
-		hwiommu_sync_single_for_device(dev, addr, size, dir);
-}
-
-void
-hwsw_sync_sg_for_device (struct device *dev, struct scatterlist *sg, int nelems, int dir)
-{
-	if (use_swiotlb(dev))
-		swiotlb_sync_sg_for_device(dev, sg, nelems, dir);
-	else
-		hwiommu_sync_sg_for_device(dev, sg, nelems, dir);
-}
-
-int
-hwsw_dma_supported (struct device *dev, u64 mask)
-{
-	if (hwiommu_dma_supported(dev, mask))
-		return 1;
-	return swiotlb_dma_supported(dev, mask);
-}
-
-int
-hwsw_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
-{
-	return hwiommu_dma_mapping_error(dev, dma_addr) ||
-		swiotlb_dma_mapping_error(dev, dma_addr);
-}
-
-EXPORT_SYMBOL(hwsw_dma_mapping_error);
-EXPORT_SYMBOL(hwsw_dma_supported);
-EXPORT_SYMBOL(hwsw_alloc_coherent);
-EXPORT_SYMBOL(hwsw_free_coherent);
-EXPORT_SYMBOL(hwsw_sync_single_for_cpu);
-EXPORT_SYMBOL(hwsw_sync_single_for_device);
-EXPORT_SYMBOL(hwsw_sync_sg_for_cpu);
-EXPORT_SYMBOL(hwsw_sync_sg_for_device);
diff --git a/arch/ia64/hp/common/sba_iommu.c b/arch/ia64/hp/common/sba_iommu.c
index 6d5e6c5630e3..56ceb68eb99d 100644
--- a/arch/ia64/hp/common/sba_iommu.c
+++ b/arch/ia64/hp/common/sba_iommu.c
@@ -36,6 +36,7 @@
 #include <linux/bitops.h>         /* hweight64() */
 #include <linux/crash_dump.h>
 #include <linux/iommu-helper.h>
+#include <linux/dma-mapping.h>
 
 #include <asm/delay.h>		/* ia64_get_itc() */
 #include <asm/io.h>
@@ -908,11 +909,13 @@ sba_mark_invalid(struct ioc *ioc, dma_addr_t iova, size_t byte_cnt)
  *
  * See Documentation/PCI/PCI-DMA-mapping.txt
  */
-dma_addr_t
-sba_map_single_attrs(struct device *dev, void *addr, size_t size, int dir,
-		     struct dma_attrs *attrs)
+static dma_addr_t sba_map_page(struct device *dev, struct page *page,
+			       unsigned long poff, size_t size,
+			       enum dma_data_direction dir,
+			       struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
+	void *addr = page_address(page) + poff;
 	dma_addr_t iovp;
 	dma_addr_t offset;
 	u64 *pdir_start;
@@ -990,7 +993,14 @@ sba_map_single_attrs(struct device *dev, void *addr, size_t size, int dir,
 #endif
 	return SBA_IOVA(ioc, iovp, offset);
 }
-EXPORT_SYMBOL(sba_map_single_attrs);
+
+static dma_addr_t sba_map_single_attrs(struct device *dev, void *addr,
+				       size_t size, enum dma_data_direction dir,
+				       struct dma_attrs *attrs)
+{
+	return sba_map_page(dev, virt_to_page(addr),
+			    (unsigned long)addr & ~PAGE_MASK, size, dir, attrs);
+}
 
 #ifdef ENABLE_MARK_CLEAN
 static SBA_INLINE void
@@ -1026,8 +1036,8 @@ sba_mark_clean(struct ioc *ioc, dma_addr_t iova, size_t size)
  *
  * See Documentation/PCI/PCI-DMA-mapping.txt
  */
-void sba_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
-			    int dir, struct dma_attrs *attrs)
+static void sba_unmap_page(struct device *dev, dma_addr_t iova, size_t size,
+			   enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 #if DELAYED_RESOURCE_CNT > 0
@@ -1094,7 +1104,12 @@ void sba_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
 	spin_unlock_irqrestore(&ioc->res_lock, flags);
 #endif /* DELAYED_RESOURCE_CNT == 0 */
 }
-EXPORT_SYMBOL(sba_unmap_single_attrs);
+
+void sba_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
+			    enum dma_data_direction dir, struct dma_attrs *attrs)
+{
+	sba_unmap_page(dev, iova, size, dir, attrs);
+}
 
 /**
  * sba_alloc_coherent - allocate/map shared mem for DMA
@@ -1104,7 +1119,7 @@ EXPORT_SYMBOL(sba_unmap_single_attrs);
  *
  * See Documentation/PCI/PCI-DMA-mapping.txt
  */
-void *
+static void *
 sba_alloc_coherent (struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t flags)
 {
 	struct ioc *ioc;
@@ -1167,7 +1182,8 @@ sba_alloc_coherent (struct device *dev, size_t size, dma_addr_t *dma_handle, gfp
  *
  * See Documentation/PCI/PCI-DMA-mapping.txt
  */
-void sba_free_coherent (struct device *dev, size_t size, void *vaddr, dma_addr_t dma_handle)
+static void sba_free_coherent (struct device *dev, size_t size, void *vaddr,
+			       dma_addr_t dma_handle)
 {
 	sba_unmap_single_attrs(dev, dma_handle, size, 0, NULL);
 	free_pages((unsigned long) vaddr, get_order(size));
@@ -1422,8 +1438,9 @@ sba_coalesce_chunks(struct ioc *ioc, struct device *dev,
  *
  * See Documentation/PCI/PCI-DMA-mapping.txt
  */
-int sba_map_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
-		     int dir, struct dma_attrs *attrs)
+static int sba_map_sg_attrs(struct device *dev, struct scatterlist *sglist,
+			    int nents, enum dma_data_direction dir,
+			    struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 	int coalesced, filled = 0;
@@ -1502,7 +1519,6 @@ int sba_map_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
 
 	return filled;
 }
-EXPORT_SYMBOL(sba_map_sg_attrs);
 
 /**
  * sba_unmap_sg_attrs - unmap Scatter/Gather list
@@ -1514,8 +1530,9 @@ EXPORT_SYMBOL(sba_map_sg_attrs);
  *
  * See Documentation/PCI/PCI-DMA-mapping.txt
  */
-void sba_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
-			int nents, int dir, struct dma_attrs *attrs)
+static void sba_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
+			       int nents, enum dma_data_direction dir,
+			       struct dma_attrs *attrs)
 {
 #ifdef ASSERT_PDIR_SANITY
 	struct ioc *ioc;
@@ -1551,7 +1568,6 @@ void sba_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
 #endif
 
 }
-EXPORT_SYMBOL(sba_unmap_sg_attrs);
 
 /**************************************************************
 *
@@ -2064,6 +2080,8 @@ static struct acpi_driver acpi_sba_ioc_driver = {
 	},
 };
 
+extern struct dma_map_ops swiotlb_dma_ops;
+
 static int __init
 sba_init(void)
 {
@@ -2077,6 +2095,7 @@ sba_init(void)
 	 * a successful kdump kernel boot is to use the swiotlb.
 	 */
 	if (is_kdump_kernel()) {
+		dma_ops = &swiotlb_dma_ops;
 		if (swiotlb_late_init_with_default_size(64 * (1<<20)) != 0)
 			panic("Unable to initialize software I/O TLB:"
 				  " Try machvec=dig boot option");
@@ -2092,6 +2111,7 @@ sba_init(void)
 		 * If we didn't find something sba_iommu can claim, we
 		 * need to setup the swiotlb and switch to the dig machvec.
 		 */
+		dma_ops = &swiotlb_dma_ops;
 		if (swiotlb_late_init_with_default_size(64 * (1<<20)) != 0)
 			panic("Unable to find SBA IOMMU or initialize "
 			      "software I/O TLB: Try machvec=dig boot option");
@@ -2138,15 +2158,13 @@ nosbagart(char *str)
 	return 1;
 }
 
-int
-sba_dma_supported (struct device *dev, u64 mask)
+static int sba_dma_supported (struct device *dev, u64 mask)
 {
 	/* make sure it's at least 32bit capable */
 	return ((mask & 0xFFFFFFFFUL) == 0xFFFFFFFFUL);
 }
 
-int
-sba_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
+static int sba_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
 {
 	return 0;
 }
@@ -2176,7 +2194,22 @@ sba_page_override(char *str)
 
 __setup("sbapagesize=",sba_page_override);
 
-EXPORT_SYMBOL(sba_dma_mapping_error);
-EXPORT_SYMBOL(sba_dma_supported);
-EXPORT_SYMBOL(sba_alloc_coherent);
-EXPORT_SYMBOL(sba_free_coherent);
+struct dma_map_ops sba_dma_ops = {
+	.alloc_coherent		= sba_alloc_coherent,
+	.free_coherent		= sba_free_coherent,
+	.map_page		= sba_map_page,
+	.unmap_page		= sba_unmap_page,
+	.map_sg			= sba_map_sg_attrs,
+	.unmap_sg		= sba_unmap_sg_attrs,
+	.sync_single_for_cpu	= machvec_dma_sync_single,
+	.sync_sg_for_cpu	= machvec_dma_sync_sg,
+	.sync_single_for_device	= machvec_dma_sync_single,
+	.sync_sg_for_device	= machvec_dma_sync_sg,
+	.dma_supported		= sba_dma_supported,
+	.mapping_error		= sba_dma_mapping_error,
+};
+
+void sba_dma_init(void)
+{
+	dma_ops = &sba_dma_ops;
+}
diff --git a/arch/ia64/include/asm/dma-mapping.h b/arch/ia64/include/asm/dma-mapping.h
index 1f912d927585..36c0009dbece 100644
--- a/arch/ia64/include/asm/dma-mapping.h
+++ b/arch/ia64/include/asm/dma-mapping.h
@@ -11,99 +11,128 @@
 
 #define ARCH_HAS_DMA_GET_REQUIRED_MASK
 
-struct dma_mapping_ops {
-	int             (*mapping_error)(struct device *dev,
-					 dma_addr_t dma_addr);
-	void*           (*alloc_coherent)(struct device *dev, size_t size,
-				dma_addr_t *dma_handle, gfp_t gfp);
-	void            (*free_coherent)(struct device *dev, size_t size,
-				void *vaddr, dma_addr_t dma_handle);
-	dma_addr_t      (*map_single)(struct device *hwdev, unsigned long ptr,
-				size_t size, int direction);
-	void            (*unmap_single)(struct device *dev, dma_addr_t addr,
-				size_t size, int direction);
-	void            (*sync_single_for_cpu)(struct device *hwdev,
-				dma_addr_t dma_handle, size_t size,
-				int direction);
-	void            (*sync_single_for_device)(struct device *hwdev,
-				dma_addr_t dma_handle, size_t size,
-				int direction);
-	void            (*sync_single_range_for_cpu)(struct device *hwdev,
-				dma_addr_t dma_handle, unsigned long offset,
-				size_t size, int direction);
-	void            (*sync_single_range_for_device)(struct device *hwdev,
-				dma_addr_t dma_handle, unsigned long offset,
-				size_t size, int direction);
-	void            (*sync_sg_for_cpu)(struct device *hwdev,
-				struct scatterlist *sg, int nelems,
-				int direction);
-	void            (*sync_sg_for_device)(struct device *hwdev,
-				struct scatterlist *sg, int nelems,
-				int direction);
-	int             (*map_sg)(struct device *hwdev, struct scatterlist *sg,
-				int nents, int direction);
-	void            (*unmap_sg)(struct device *hwdev,
-				struct scatterlist *sg, int nents,
-				int direction);
-	int             (*dma_supported_op)(struct device *hwdev, u64 mask);
-	int		is_phys;
-};
-
-extern struct dma_mapping_ops *dma_ops;
+extern struct dma_map_ops *dma_ops;
 extern struct ia64_machine_vector ia64_mv;
 extern void set_iommu_machvec(void);
 
-#define dma_alloc_coherent(dev, size, handle, gfp)	\
-	platform_dma_alloc_coherent(dev, size, handle, (gfp) | GFP_DMA)
+extern void machvec_dma_sync_single(struct device *, dma_addr_t, size_t,
+				    enum dma_data_direction);
+extern void machvec_dma_sync_sg(struct device *, struct scatterlist *, int,
+				enum dma_data_direction);
 
-/* coherent mem. is cheap */
-static inline void *
-dma_alloc_noncoherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
-		      gfp_t flag)
+static inline void *dma_alloc_coherent(struct device *dev, size_t size,
+				       dma_addr_t *daddr, gfp_t gfp)
 {
-	return dma_alloc_coherent(dev, size, dma_handle, flag);
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	return ops->alloc_coherent(dev, size, daddr, gfp);
 }
-#define dma_free_coherent	platform_dma_free_coherent
-static inline void
-dma_free_noncoherent(struct device *dev, size_t size, void *cpu_addr,
-		     dma_addr_t dma_handle)
+
+static inline void dma_free_coherent(struct device *dev, size_t size,
+				     void *caddr, dma_addr_t daddr)
+{
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	ops->free_coherent(dev, size, caddr, daddr);
+}
+
+#define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f)
+#define dma_free_noncoherent(d, s, v, h) dma_free_coherent(d, s, v, h)
+
+static inline dma_addr_t dma_map_single_attrs(struct device *dev,
+					      void *caddr, size_t size,
+					      enum dma_data_direction dir,
+					      struct dma_attrs *attrs)
+{
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	return ops->map_page(dev, virt_to_page(caddr),
+			     (unsigned long)caddr & ~PAGE_MASK, size,
+			     dir, attrs);
+}
+
+static inline void dma_unmap_single_attrs(struct device *dev, dma_addr_t daddr,
+					  size_t size,
+					  enum dma_data_direction dir,
+					  struct dma_attrs *attrs)
+{
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	ops->unmap_page(dev, daddr, size, dir, attrs);
+}
+
+#define dma_map_single(d, a, s, r) dma_map_single_attrs(d, a, s, r, NULL)
+#define dma_unmap_single(d, a, s, r) dma_unmap_single_attrs(d, a, s, r, NULL)
+
+static inline int dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
+				   int nents, enum dma_data_direction dir,
+				   struct dma_attrs *attrs)
+{
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	return ops->map_sg(dev, sgl, nents, dir, attrs);
+}
+
+static inline void dma_unmap_sg_attrs(struct device *dev,
+				      struct scatterlist *sgl, int nents,
+				      enum dma_data_direction dir,
+				      struct dma_attrs *attrs)
+{
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	ops->unmap_sg(dev, sgl, nents, dir, attrs);
+}
+
+#define dma_map_sg(d, s, n, r) dma_map_sg_attrs(d, s, n, r, NULL)
+#define dma_unmap_sg(d, s, n, r) dma_unmap_sg_attrs(d, s, n, r, NULL)
+
+static inline void dma_sync_single_for_cpu(struct device *dev, dma_addr_t daddr,
+					   size_t size,
+					   enum dma_data_direction dir)
 {
-	dma_free_coherent(dev, size, cpu_addr, dma_handle);
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	ops->sync_single_for_cpu(dev, daddr, size, dir);
 }
-#define dma_map_single_attrs	platform_dma_map_single_attrs
-static inline dma_addr_t dma_map_single(struct device *dev, void *cpu_addr,
-					size_t size, int dir)
+
+static inline void dma_sync_sg_for_cpu(struct device *dev,
+				       struct scatterlist *sgl,
+				       int nents, enum dma_data_direction dir)
 {
-	return dma_map_single_attrs(dev, cpu_addr, size, dir, NULL);
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	ops->sync_sg_for_cpu(dev, sgl, nents, dir);
 }
-#define dma_map_sg_attrs	platform_dma_map_sg_attrs
-static inline int dma_map_sg(struct device *dev, struct scatterlist *sgl,
-			     int nents, int dir)
+
+static inline void dma_sync_single_for_device(struct device *dev,
+					      dma_addr_t daddr,
+					      size_t size,
+					      enum dma_data_direction dir)
 {
-	return dma_map_sg_attrs(dev, sgl, nents, dir, NULL);
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	ops->sync_single_for_device(dev, daddr, size, dir);
 }
-#define dma_unmap_single_attrs	platform_dma_unmap_single_attrs
-static inline void dma_unmap_single(struct device *dev, dma_addr_t cpu_addr,
-				    size_t size, int dir)
+
+static inline void dma_sync_sg_for_device(struct device *dev,
+					  struct scatterlist *sgl,
+					  int nents,
+					  enum dma_data_direction dir)
 {
-	return dma_unmap_single_attrs(dev, cpu_addr, size, dir, NULL);
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	ops->sync_sg_for_device(dev, sgl, nents, dir);
 }
-#define dma_unmap_sg_attrs	platform_dma_unmap_sg_attrs
-static inline void dma_unmap_sg(struct device *dev, struct scatterlist *sgl,
-				int nents, int dir)
+
+static inline int dma_mapping_error(struct device *dev, dma_addr_t daddr)
+{
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	return ops->mapping_error(dev, daddr);
+}
+
+static inline dma_addr_t dma_map_page(struct device *dev, struct page *page,
+				      size_t offset, size_t size,
+				      enum dma_data_direction dir)
 {
-	return dma_unmap_sg_attrs(dev, sgl, nents, dir, NULL);
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	return ops->map_page(dev, page, offset, size, dir, NULL);
 }
-#define dma_sync_single_for_cpu	platform_dma_sync_single_for_cpu
-#define dma_sync_sg_for_cpu	platform_dma_sync_sg_for_cpu
-#define dma_sync_single_for_device platform_dma_sync_single_for_device
-#define dma_sync_sg_for_device	platform_dma_sync_sg_for_device
-#define dma_mapping_error	platform_dma_mapping_error
 
-#define dma_map_page(dev, pg, off, size, dir)				\
-	dma_map_single(dev, page_address(pg) + (off), (size), (dir))
-#define dma_unmap_page(dev, dma_addr, size, dir)			\
-	dma_unmap_single(dev, dma_addr, size, dir)
+static inline void dma_unmap_page(struct device *dev, dma_addr_t addr,
+				  size_t size, enum dma_data_direction dir)
+{
+	dma_unmap_single(dev, addr, size, dir);
+}
 
 /*
  * Rest of this file is part of the "Advanced DMA API".  Use at your own risk.
@@ -115,7 +144,11 @@ static inline void dma_unmap_sg(struct device *dev, struct scatterlist *sgl,
 #define dma_sync_single_range_for_device(dev, dma_handle, offset, size, dir)	\
 	dma_sync_single_for_device(dev, dma_handle, size, dir)
 
-#define dma_supported		platform_dma_supported
+static inline int dma_supported(struct device *dev, u64 mask)
+{
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
+	return ops->dma_supported(dev, mask);
+}
 
 static inline int
 dma_set_mask (struct device *dev, u64 mask)
@@ -141,11 +174,4 @@ dma_cache_sync (struct device *dev, void *vaddr, size_t size,
 
 #define dma_is_consistent(d, h)	(1)	/* all we do is coherent memory... */
 
-static inline struct dma_mapping_ops *get_dma_ops(struct device *dev)
-{
-	return dma_ops;
-}
-
-
-
 #endif /* _ASM_IA64_DMA_MAPPING_H */
diff --git a/arch/ia64/include/asm/machvec.h b/arch/ia64/include/asm/machvec.h
index fe87b2121707..367d299d9938 100644
--- a/arch/ia64/include/asm/machvec.h
+++ b/arch/ia64/include/asm/machvec.h
@@ -11,7 +11,6 @@
 #define _ASM_IA64_MACHVEC_H
 
 #include <linux/types.h>
-#include <linux/swiotlb.h>
 
 /* forward declarations: */
 struct device;
@@ -45,24 +44,8 @@ typedef void ia64_mv_kernel_launch_event_t(void);
 
 /* DMA-mapping interface: */
 typedef void ia64_mv_dma_init (void);
-typedef void *ia64_mv_dma_alloc_coherent (struct device *, size_t, dma_addr_t *, gfp_t);
-typedef void ia64_mv_dma_free_coherent (struct device *, size_t, void *, dma_addr_t);
-typedef dma_addr_t ia64_mv_dma_map_single (struct device *, void *, size_t, int);
-typedef void ia64_mv_dma_unmap_single (struct device *, dma_addr_t, size_t, int);
-typedef int ia64_mv_dma_map_sg (struct device *, struct scatterlist *, int, int);
-typedef void ia64_mv_dma_unmap_sg (struct device *, struct scatterlist *, int, int);
-typedef void ia64_mv_dma_sync_single_for_cpu (struct device *, dma_addr_t, size_t, int);
-typedef void ia64_mv_dma_sync_sg_for_cpu (struct device *, struct scatterlist *, int, int);
-typedef void ia64_mv_dma_sync_single_for_device (struct device *, dma_addr_t, size_t, int);
-typedef void ia64_mv_dma_sync_sg_for_device (struct device *, struct scatterlist *, int, int);
-typedef int ia64_mv_dma_mapping_error(struct device *, dma_addr_t dma_addr);
-typedef int ia64_mv_dma_supported (struct device *, u64);
-
-typedef dma_addr_t ia64_mv_dma_map_single_attrs (struct device *, void *, size_t, int, struct dma_attrs *);
-typedef void ia64_mv_dma_unmap_single_attrs (struct device *, dma_addr_t, size_t, int, struct dma_attrs *);
-typedef int ia64_mv_dma_map_sg_attrs (struct device *, struct scatterlist *, int, int, struct dma_attrs *);
-typedef void ia64_mv_dma_unmap_sg_attrs (struct device *, struct scatterlist *, int, int, struct dma_attrs *);
 typedef u64 ia64_mv_dma_get_required_mask (struct device *);
+typedef struct dma_map_ops *ia64_mv_dma_get_ops(struct device *);
 
 /*
  * WARNING: The legacy I/O space is _architected_.  Platforms are
@@ -114,8 +97,6 @@ machvec_noop_bus (struct pci_bus *bus)
 
 extern void machvec_setup (char **);
 extern void machvec_timer_interrupt (int, void *);
-extern void machvec_dma_sync_single (struct device *, dma_addr_t, size_t, int);
-extern void machvec_dma_sync_sg (struct device *, struct scatterlist *, int, int);
 extern void machvec_tlb_migrate_finish (struct mm_struct *);
 
 # if defined (CONFIG_IA64_HP_SIM)
@@ -148,19 +129,8 @@ extern void machvec_tlb_migrate_finish (struct mm_struct *);
 #  define platform_global_tlb_purge	ia64_mv.global_tlb_purge
 #  define platform_tlb_migrate_finish	ia64_mv.tlb_migrate_finish
 #  define platform_dma_init		ia64_mv.dma_init
-#  define platform_dma_alloc_coherent	ia64_mv.dma_alloc_coherent
-#  define platform_dma_free_coherent	ia64_mv.dma_free_coherent
-#  define platform_dma_map_single_attrs	ia64_mv.dma_map_single_attrs
-#  define platform_dma_unmap_single_attrs	ia64_mv.dma_unmap_single_attrs
-#  define platform_dma_map_sg_attrs	ia64_mv.dma_map_sg_attrs
-#  define platform_dma_unmap_sg_attrs	ia64_mv.dma_unmap_sg_attrs
-#  define platform_dma_sync_single_for_cpu ia64_mv.dma_sync_single_for_cpu
-#  define platform_dma_sync_sg_for_cpu	ia64_mv.dma_sync_sg_for_cpu
-#  define platform_dma_sync_single_for_device ia64_mv.dma_sync_single_for_device
-#  define platform_dma_sync_sg_for_device ia64_mv.dma_sync_sg_for_device
-#  define platform_dma_mapping_error		ia64_mv.dma_mapping_error
-#  define platform_dma_supported	ia64_mv.dma_supported
 #  define platform_dma_get_required_mask ia64_mv.dma_get_required_mask
+#  define platform_dma_get_ops		ia64_mv.dma_get_ops
 #  define platform_irq_to_vector	ia64_mv.irq_to_vector
 #  define platform_local_vector_to_irq	ia64_mv.local_vector_to_irq
 #  define platform_pci_get_legacy_mem	ia64_mv.pci_get_legacy_mem
@@ -203,19 +173,8 @@ struct ia64_machine_vector {
 	ia64_mv_global_tlb_purge_t *global_tlb_purge;
 	ia64_mv_tlb_migrate_finish_t *tlb_migrate_finish;
 	ia64_mv_dma_init *dma_init;
-	ia64_mv_dma_alloc_coherent *dma_alloc_coherent;
-	ia64_mv_dma_free_coherent *dma_free_coherent;
-	ia64_mv_dma_map_single_attrs *dma_map_single_attrs;
-	ia64_mv_dma_unmap_single_attrs *dma_unmap_single_attrs;
-	ia64_mv_dma_map_sg_attrs *dma_map_sg_attrs;
-	ia64_mv_dma_unmap_sg_attrs *dma_unmap_sg_attrs;
-	ia64_mv_dma_sync_single_for_cpu *dma_sync_single_for_cpu;
-	ia64_mv_dma_sync_sg_for_cpu *dma_sync_sg_for_cpu;
-	ia64_mv_dma_sync_single_for_device *dma_sync_single_for_device;
-	ia64_mv_dma_sync_sg_for_device *dma_sync_sg_for_device;
-	ia64_mv_dma_mapping_error *dma_mapping_error;
-	ia64_mv_dma_supported *dma_supported;
 	ia64_mv_dma_get_required_mask *dma_get_required_mask;
+	ia64_mv_dma_get_ops *dma_get_ops;
 	ia64_mv_irq_to_vector *irq_to_vector;
 	ia64_mv_local_vector_to_irq *local_vector_to_irq;
 	ia64_mv_pci_get_legacy_mem_t *pci_get_legacy_mem;
@@ -254,19 +213,8 @@ struct ia64_machine_vector {
 	platform_global_tlb_purge,		\
 	platform_tlb_migrate_finish,		\
 	platform_dma_init,			\
-	platform_dma_alloc_coherent,		\
-	platform_dma_free_coherent,		\
-	platform_dma_map_single_attrs,		\
-	platform_dma_unmap_single_attrs,	\
-	platform_dma_map_sg_attrs,		\
-	platform_dma_unmap_sg_attrs,		\
-	platform_dma_sync_single_for_cpu,	\
-	platform_dma_sync_sg_for_cpu,		\
-	platform_dma_sync_single_for_device,	\
-	platform_dma_sync_sg_for_device,	\
-	platform_dma_mapping_error,			\
-	platform_dma_supported,			\
 	platform_dma_get_required_mask,		\
+	platform_dma_get_ops,			\
 	platform_irq_to_vector,			\
 	platform_local_vector_to_irq,		\
 	platform_pci_get_legacy_mem,		\
@@ -302,6 +250,9 @@ extern void machvec_init_from_cmdline(const char *cmdline);
 #  error Unknown configuration.  Update arch/ia64/include/asm/machvec.h.
 # endif /* CONFIG_IA64_GENERIC */
 
+extern void swiotlb_dma_init(void);
+extern struct dma_map_ops *dma_get_ops(struct device *);
+
 /*
  * Define default versions so we can extend machvec for new platforms without having
  * to update the machvec files for all existing platforms.
@@ -332,43 +283,10 @@ extern void machvec_init_from_cmdline(const char *cmdline);
 # define platform_kernel_launch_event	machvec_noop
 #endif
 #ifndef platform_dma_init
-# define platform_dma_init		swiotlb_init
-#endif
-#ifndef platform_dma_alloc_coherent
-# define platform_dma_alloc_coherent	swiotlb_alloc_coherent
-#endif
-#ifndef platform_dma_free_coherent
-# define platform_dma_free_coherent	swiotlb_free_coherent
-#endif
-#ifndef platform_dma_map_single_attrs
-# define platform_dma_map_single_attrs	swiotlb_map_single_attrs
-#endif
-#ifndef platform_dma_unmap_single_attrs
-# define platform_dma_unmap_single_attrs	swiotlb_unmap_single_attrs
-#endif
-#ifndef platform_dma_map_sg_attrs
-# define platform_dma_map_sg_attrs	swiotlb_map_sg_attrs
-#endif
-#ifndef platform_dma_unmap_sg_attrs
-# define platform_dma_unmap_sg_attrs	swiotlb_unmap_sg_attrs
-#endif
-#ifndef platform_dma_sync_single_for_cpu
-# define platform_dma_sync_single_for_cpu	swiotlb_sync_single_for_cpu
-#endif
-#ifndef platform_dma_sync_sg_for_cpu
-# define platform_dma_sync_sg_for_cpu		swiotlb_sync_sg_for_cpu
-#endif
-#ifndef platform_dma_sync_single_for_device
-# define platform_dma_sync_single_for_device	swiotlb_sync_single_for_device
-#endif
-#ifndef platform_dma_sync_sg_for_device
-# define platform_dma_sync_sg_for_device	swiotlb_sync_sg_for_device
-#endif
-#ifndef platform_dma_mapping_error
-# define platform_dma_mapping_error		swiotlb_dma_mapping_error
+# define platform_dma_init		swiotlb_dma_init
 #endif
-#ifndef platform_dma_supported
-# define  platform_dma_supported	swiotlb_dma_supported
+#ifndef platform_dma_get_ops
+# define platform_dma_get_ops		dma_get_ops
 #endif
 #ifndef platform_dma_get_required_mask
 # define  platform_dma_get_required_mask	ia64_dma_get_required_mask
diff --git a/arch/ia64/include/asm/machvec_dig_vtd.h b/arch/ia64/include/asm/machvec_dig_vtd.h
index 3400b561e711..6ab1de5c45ef 100644
--- a/arch/ia64/include/asm/machvec_dig_vtd.h
+++ b/arch/ia64/include/asm/machvec_dig_vtd.h
@@ -2,14 +2,6 @@
 #define _ASM_IA64_MACHVEC_DIG_VTD_h
 
 extern ia64_mv_setup_t			dig_setup;
-extern ia64_mv_dma_alloc_coherent	vtd_alloc_coherent;
-extern ia64_mv_dma_free_coherent	vtd_free_coherent;
-extern ia64_mv_dma_map_single_attrs	vtd_map_single_attrs;
-extern ia64_mv_dma_unmap_single_attrs	vtd_unmap_single_attrs;
-extern ia64_mv_dma_map_sg_attrs		vtd_map_sg_attrs;
-extern ia64_mv_dma_unmap_sg_attrs	vtd_unmap_sg_attrs;
-extern ia64_mv_dma_supported		iommu_dma_supported;
-extern ia64_mv_dma_mapping_error	vtd_dma_mapping_error;
 extern ia64_mv_dma_init			pci_iommu_alloc;
 
 /*
@@ -22,17 +14,5 @@ extern ia64_mv_dma_init			pci_iommu_alloc;
 #define platform_name				"dig_vtd"
 #define platform_setup				dig_setup
 #define platform_dma_init			pci_iommu_alloc
-#define platform_dma_alloc_coherent		vtd_alloc_coherent
-#define platform_dma_free_coherent		vtd_free_coherent
-#define platform_dma_map_single_attrs		vtd_map_single_attrs
-#define platform_dma_unmap_single_attrs		vtd_unmap_single_attrs
-#define platform_dma_map_sg_attrs		vtd_map_sg_attrs
-#define platform_dma_unmap_sg_attrs		vtd_unmap_sg_attrs
-#define platform_dma_sync_single_for_cpu	machvec_dma_sync_single
-#define platform_dma_sync_sg_for_cpu		machvec_dma_sync_sg
-#define platform_dma_sync_single_for_device	machvec_dma_sync_single
-#define platform_dma_sync_sg_for_device		machvec_dma_sync_sg
-#define platform_dma_supported			iommu_dma_supported
-#define platform_dma_mapping_error		vtd_dma_mapping_error
 
 #endif /* _ASM_IA64_MACHVEC_DIG_VTD_h */
diff --git a/arch/ia64/include/asm/machvec_hpzx1.h b/arch/ia64/include/asm/machvec_hpzx1.h
index 2f57f5144b9f..3bd83d78a412 100644
--- a/arch/ia64/include/asm/machvec_hpzx1.h
+++ b/arch/ia64/include/asm/machvec_hpzx1.h
@@ -2,14 +2,7 @@
 #define _ASM_IA64_MACHVEC_HPZX1_h
 
 extern ia64_mv_setup_t			dig_setup;
-extern ia64_mv_dma_alloc_coherent	sba_alloc_coherent;
-extern ia64_mv_dma_free_coherent	sba_free_coherent;
-extern ia64_mv_dma_map_single_attrs	sba_map_single_attrs;
-extern ia64_mv_dma_unmap_single_attrs	sba_unmap_single_attrs;
-extern ia64_mv_dma_map_sg_attrs		sba_map_sg_attrs;
-extern ia64_mv_dma_unmap_sg_attrs	sba_unmap_sg_attrs;
-extern ia64_mv_dma_supported		sba_dma_supported;
-extern ia64_mv_dma_mapping_error	sba_dma_mapping_error;
+extern ia64_mv_dma_init			sba_dma_init;
 
 /*
  * This stuff has dual use!
@@ -20,18 +13,6 @@ extern ia64_mv_dma_mapping_error	sba_dma_mapping_error;
  */
 #define platform_name				"hpzx1"
 #define platform_setup				dig_setup
-#define platform_dma_init			machvec_noop
-#define platform_dma_alloc_coherent		sba_alloc_coherent
-#define platform_dma_free_coherent		sba_free_coherent
-#define platform_dma_map_single_attrs		sba_map_single_attrs
-#define platform_dma_unmap_single_attrs		sba_unmap_single_attrs
-#define platform_dma_map_sg_attrs		sba_map_sg_attrs
-#define platform_dma_unmap_sg_attrs		sba_unmap_sg_attrs
-#define platform_dma_sync_single_for_cpu	machvec_dma_sync_single
-#define platform_dma_sync_sg_for_cpu		machvec_dma_sync_sg
-#define platform_dma_sync_single_for_device	machvec_dma_sync_single
-#define platform_dma_sync_sg_for_device		machvec_dma_sync_sg
-#define platform_dma_supported			sba_dma_supported
-#define platform_dma_mapping_error		sba_dma_mapping_error
+#define platform_dma_init			sba_dma_init
 
 #endif /* _ASM_IA64_MACHVEC_HPZX1_h */
diff --git a/arch/ia64/include/asm/machvec_hpzx1_swiotlb.h b/arch/ia64/include/asm/machvec_hpzx1_swiotlb.h
index a842cdda827b..1091ac39740c 100644
--- a/arch/ia64/include/asm/machvec_hpzx1_swiotlb.h
+++ b/arch/ia64/include/asm/machvec_hpzx1_swiotlb.h
@@ -2,18 +2,7 @@
 #define _ASM_IA64_MACHVEC_HPZX1_SWIOTLB_h
 
 extern ia64_mv_setup_t				dig_setup;
-extern ia64_mv_dma_alloc_coherent		hwsw_alloc_coherent;
-extern ia64_mv_dma_free_coherent		hwsw_free_coherent;
-extern ia64_mv_dma_map_single_attrs		hwsw_map_single_attrs;
-extern ia64_mv_dma_unmap_single_attrs		hwsw_unmap_single_attrs;
-extern ia64_mv_dma_map_sg_attrs			hwsw_map_sg_attrs;
-extern ia64_mv_dma_unmap_sg_attrs		hwsw_unmap_sg_attrs;
-extern ia64_mv_dma_supported			hwsw_dma_supported;
-extern ia64_mv_dma_mapping_error		hwsw_dma_mapping_error;
-extern ia64_mv_dma_sync_single_for_cpu		hwsw_sync_single_for_cpu;
-extern ia64_mv_dma_sync_sg_for_cpu		hwsw_sync_sg_for_cpu;
-extern ia64_mv_dma_sync_single_for_device	hwsw_sync_single_for_device;
-extern ia64_mv_dma_sync_sg_for_device		hwsw_sync_sg_for_device;
+extern ia64_mv_dma_get_ops			hwsw_dma_get_ops;
 
 /*
  * This stuff has dual use!
@@ -23,20 +12,8 @@ extern ia64_mv_dma_sync_sg_for_device		hwsw_sync_sg_for_device;
  * the macros are used directly.
  */
 #define platform_name				"hpzx1_swiotlb"
-
 #define platform_setup				dig_setup
 #define platform_dma_init			machvec_noop
-#define platform_dma_alloc_coherent		hwsw_alloc_coherent
-#define platform_dma_free_coherent		hwsw_free_coherent
-#define platform_dma_map_single_attrs		hwsw_map_single_attrs
-#define platform_dma_unmap_single_attrs		hwsw_unmap_single_attrs
-#define platform_dma_map_sg_attrs		hwsw_map_sg_attrs
-#define platform_dma_unmap_sg_attrs		hwsw_unmap_sg_attrs
-#define platform_dma_supported			hwsw_dma_supported
-#define platform_dma_mapping_error		hwsw_dma_mapping_error
-#define platform_dma_sync_single_for_cpu	hwsw_sync_single_for_cpu
-#define platform_dma_sync_sg_for_cpu		hwsw_sync_sg_for_cpu
-#define platform_dma_sync_single_for_device	hwsw_sync_single_for_device
-#define platform_dma_sync_sg_for_device		hwsw_sync_sg_for_device
+#define platform_dma_get_ops			hwsw_dma_get_ops
 
 #endif /* _ASM_IA64_MACHVEC_HPZX1_SWIOTLB_h */
diff --git a/arch/ia64/include/asm/machvec_sn2.h b/arch/ia64/include/asm/machvec_sn2.h
index f1a6e0d6dfa5..f061a30aac42 100644
--- a/arch/ia64/include/asm/machvec_sn2.h
+++ b/arch/ia64/include/asm/machvec_sn2.h
@@ -55,19 +55,8 @@ extern ia64_mv_readb_t __sn_readb_relaxed;
 extern ia64_mv_readw_t __sn_readw_relaxed;
 extern ia64_mv_readl_t __sn_readl_relaxed;
 extern ia64_mv_readq_t __sn_readq_relaxed;
-extern ia64_mv_dma_alloc_coherent	sn_dma_alloc_coherent;
-extern ia64_mv_dma_free_coherent	sn_dma_free_coherent;
-extern ia64_mv_dma_map_single_attrs	sn_dma_map_single_attrs;
-extern ia64_mv_dma_unmap_single_attrs	sn_dma_unmap_single_attrs;
-extern ia64_mv_dma_map_sg_attrs		sn_dma_map_sg_attrs;
-extern ia64_mv_dma_unmap_sg_attrs	sn_dma_unmap_sg_attrs;
-extern ia64_mv_dma_sync_single_for_cpu	sn_dma_sync_single_for_cpu;
-extern ia64_mv_dma_sync_sg_for_cpu	sn_dma_sync_sg_for_cpu;
-extern ia64_mv_dma_sync_single_for_device sn_dma_sync_single_for_device;
-extern ia64_mv_dma_sync_sg_for_device	sn_dma_sync_sg_for_device;
-extern ia64_mv_dma_mapping_error	sn_dma_mapping_error;
-extern ia64_mv_dma_supported		sn_dma_supported;
 extern ia64_mv_dma_get_required_mask	sn_dma_get_required_mask;
+extern ia64_mv_dma_init			sn_dma_init;
 extern ia64_mv_migrate_t		sn_migrate;
 extern ia64_mv_kernel_launch_event_t	sn_kernel_launch_event;
 extern ia64_mv_setup_msi_irq_t		sn_setup_msi_irq;
@@ -111,20 +100,8 @@ extern ia64_mv_pci_fixup_bus_t		sn_pci_fixup_bus;
 #define platform_pci_get_legacy_mem	sn_pci_get_legacy_mem
 #define platform_pci_legacy_read	sn_pci_legacy_read
 #define platform_pci_legacy_write	sn_pci_legacy_write
-#define platform_dma_init		machvec_noop
-#define platform_dma_alloc_coherent	sn_dma_alloc_coherent
-#define platform_dma_free_coherent	sn_dma_free_coherent
-#define platform_dma_map_single_attrs	sn_dma_map_single_attrs
-#define platform_dma_unmap_single_attrs	sn_dma_unmap_single_attrs
-#define platform_dma_map_sg_attrs	sn_dma_map_sg_attrs
-#define platform_dma_unmap_sg_attrs	sn_dma_unmap_sg_attrs
-#define platform_dma_sync_single_for_cpu sn_dma_sync_single_for_cpu
-#define platform_dma_sync_sg_for_cpu	sn_dma_sync_sg_for_cpu
-#define platform_dma_sync_single_for_device sn_dma_sync_single_for_device
-#define platform_dma_sync_sg_for_device	sn_dma_sync_sg_for_device
-#define platform_dma_mapping_error		sn_dma_mapping_error
-#define platform_dma_supported		sn_dma_supported
 #define platform_dma_get_required_mask	sn_dma_get_required_mask
+#define platform_dma_init		sn_dma_init
 #define platform_migrate		sn_migrate
 #define platform_kernel_launch_event    sn_kernel_launch_event
 #ifdef CONFIG_PCI_MSI
diff --git a/arch/ia64/kernel/Makefile b/arch/ia64/kernel/Makefile
index c381ea954892..f2778f2c4fd9 100644
--- a/arch/ia64/kernel/Makefile
+++ b/arch/ia64/kernel/Makefile
@@ -7,7 +7,7 @@ extra-y	:= head.o init_task.o vmlinux.lds
 obj-y := acpi.o entry.o efi.o efi_stub.o gate-data.o fsys.o ia64_ksyms.o irq.o irq_ia64.o	\
 	 irq_lsapic.o ivt.o machvec.o pal.o patch.o process.o perfmon.o ptrace.o sal.o		\
 	 salinfo.o setup.o signal.o sys_ia64.o time.o traps.o unaligned.o \
-	 unwind.o mca.o mca_asm.o topology.o
+	 unwind.o mca.o mca_asm.o topology.o dma-mapping.o
 
 obj-$(CONFIG_IA64_BRL_EMU)	+= brl_emu.o
 obj-$(CONFIG_IA64_GENERIC)	+= acpi-ext.o
@@ -43,9 +43,7 @@ ifneq ($(CONFIG_IA64_ESI),)
 obj-y				+= esi_stub.o	# must be in kernel proper
 endif
 obj-$(CONFIG_DMAR)		+= pci-dma.o
-ifeq ($(CONFIG_DMAR), y)
 obj-$(CONFIG_SWIOTLB)		+= pci-swiotlb.o
-endif
 
 # The gate DSO image is built using a special linker script.
 targets += gate.so gate-syms.o
diff --git a/arch/ia64/kernel/dma-mapping.c b/arch/ia64/kernel/dma-mapping.c
new file mode 100644
index 000000000000..086a2aeb0404
--- /dev/null
+++ b/arch/ia64/kernel/dma-mapping.c
@@ -0,0 +1,13 @@
+#include <linux/dma-mapping.h>
+
+/* Set this to 1 if there is a HW IOMMU in the system */
+int iommu_detected __read_mostly;
+
+struct dma_map_ops *dma_ops;
+EXPORT_SYMBOL(dma_ops);
+
+struct dma_map_ops *dma_get_ops(struct device *dev)
+{
+	return dma_ops;
+}
+EXPORT_SYMBOL(dma_get_ops);
diff --git a/arch/ia64/kernel/machvec.c b/arch/ia64/kernel/machvec.c
index 7ccb228ceedc..d41a40ef80c0 100644
--- a/arch/ia64/kernel/machvec.c
+++ b/arch/ia64/kernel/machvec.c
@@ -1,5 +1,5 @@
 #include <linux/module.h>
-
+#include <linux/dma-mapping.h>
 #include <asm/machvec.h>
 #include <asm/system.h>
 
@@ -75,14 +75,16 @@ machvec_timer_interrupt (int irq, void *dev_id)
 EXPORT_SYMBOL(machvec_timer_interrupt);
 
 void
-machvec_dma_sync_single (struct device *hwdev, dma_addr_t dma_handle, size_t size, int dir)
+machvec_dma_sync_single(struct device *hwdev, dma_addr_t dma_handle, size_t size,
+			enum dma_data_direction dir)
 {
 	mb();
 }
 EXPORT_SYMBOL(machvec_dma_sync_single);
 
 void
-machvec_dma_sync_sg (struct device *hwdev, struct scatterlist *sg, int n, int dir)
+machvec_dma_sync_sg(struct device *hwdev, struct scatterlist *sg, int n,
+		    enum dma_data_direction dir)
 {
 	mb();
 }
diff --git a/arch/ia64/kernel/pci-dma.c b/arch/ia64/kernel/pci-dma.c
index d0ada067a4af..e4cb443bb988 100644
--- a/arch/ia64/kernel/pci-dma.c
+++ b/arch/ia64/kernel/pci-dma.c
@@ -32,9 +32,6 @@ int force_iommu __read_mostly = 1;
 int force_iommu __read_mostly;
 #endif
 
-/* Set this to 1 if there is a HW IOMMU in the system */
-int iommu_detected __read_mostly;
-
 /* Dummy device used for NULL arguments (normally ISA). Better would
    be probably a smaller DMA mask, but this is bug-to-bug compatible
    to i386. */
@@ -44,18 +41,7 @@ struct device fallback_dev = {
 	.dma_mask = &fallback_dev.coherent_dma_mask,
 };
 
-void __init pci_iommu_alloc(void)
-{
-	/*
-	 * The order of these functions is important for
-	 * fall-back/fail-over reasons
-	 */
-	detect_intel_iommu();
-
-#ifdef CONFIG_SWIOTLB
-	pci_swiotlb_init();
-#endif
-}
+extern struct dma_map_ops intel_dma_ops;
 
 static int __init pci_iommu_init(void)
 {
@@ -79,15 +65,12 @@ iommu_dma_init(void)
 	return;
 }
 
-struct dma_mapping_ops *dma_ops;
-EXPORT_SYMBOL(dma_ops);
-
 int iommu_dma_supported(struct device *dev, u64 mask)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(dev);
+	struct dma_map_ops *ops = platform_dma_get_ops(dev);
 
-	if (ops->dma_supported_op)
-		return ops->dma_supported_op(dev, mask);
+	if (ops->dma_supported)
+		return ops->dma_supported(dev, mask);
 
 	/* Copied from i386. Doesn't make much sense, because it will
 	   only work for pci_alloc_coherent.
@@ -116,4 +99,25 @@ int iommu_dma_supported(struct device *dev, u64 mask)
 }
 EXPORT_SYMBOL(iommu_dma_supported);
 
+void __init pci_iommu_alloc(void)
+{
+	dma_ops = &intel_dma_ops;
+
+	dma_ops->sync_single_for_cpu = machvec_dma_sync_single;
+	dma_ops->sync_sg_for_cpu = machvec_dma_sync_sg;
+	dma_ops->sync_single_for_device = machvec_dma_sync_single;
+	dma_ops->sync_sg_for_device = machvec_dma_sync_sg;
+	dma_ops->dma_supported = iommu_dma_supported;
+
+	/*
+	 * The order of these functions is important for
+	 * fall-back/fail-over reasons
+	 */
+	detect_intel_iommu();
+
+#ifdef CONFIG_SWIOTLB
+	pci_swiotlb_init();
+#endif
+}
+
 #endif
diff --git a/arch/ia64/kernel/pci-swiotlb.c b/arch/ia64/kernel/pci-swiotlb.c
index 16c50516dbc1..573f02c39a00 100644
--- a/arch/ia64/kernel/pci-swiotlb.c
+++ b/arch/ia64/kernel/pci-swiotlb.c
@@ -13,23 +13,37 @@
 int swiotlb __read_mostly;
 EXPORT_SYMBOL(swiotlb);
 
-struct dma_mapping_ops swiotlb_dma_ops = {
-	.mapping_error = swiotlb_dma_mapping_error,
-	.alloc_coherent = swiotlb_alloc_coherent,
+static void *ia64_swiotlb_alloc_coherent(struct device *dev, size_t size,
+					 dma_addr_t *dma_handle, gfp_t gfp)
+{
+	if (dev->coherent_dma_mask != DMA_64BIT_MASK)
+		gfp |= GFP_DMA;
+	return swiotlb_alloc_coherent(dev, size, dma_handle, gfp);
+}
+
+struct dma_map_ops swiotlb_dma_ops = {
+	.alloc_coherent = ia64_swiotlb_alloc_coherent,
 	.free_coherent = swiotlb_free_coherent,
-	.map_single = swiotlb_map_single,
-	.unmap_single = swiotlb_unmap_single,
+	.map_page = swiotlb_map_page,
+	.unmap_page = swiotlb_unmap_page,
+	.map_sg = swiotlb_map_sg_attrs,
+	.unmap_sg = swiotlb_unmap_sg_attrs,
 	.sync_single_for_cpu = swiotlb_sync_single_for_cpu,
 	.sync_single_for_device = swiotlb_sync_single_for_device,
 	.sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
 	.sync_single_range_for_device = swiotlb_sync_single_range_for_device,
 	.sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
 	.sync_sg_for_device = swiotlb_sync_sg_for_device,
-	.map_sg = swiotlb_map_sg,
-	.unmap_sg = swiotlb_unmap_sg,
-	.dma_supported_op = swiotlb_dma_supported,
+	.dma_supported = swiotlb_dma_supported,
+	.mapping_error = swiotlb_dma_mapping_error,
 };
 
+void __init swiotlb_dma_init(void)
+{
+	dma_ops = &swiotlb_dma_ops;
+	swiotlb_init();
+}
+
 void __init pci_swiotlb_init(void)
 {
 	if (!iommu_detected) {
diff --git a/arch/ia64/sn/pci/pci_dma.c b/arch/ia64/sn/pci/pci_dma.c
index 863f5017baae..8c130e8f00e1 100644
--- a/arch/ia64/sn/pci/pci_dma.c
+++ b/arch/ia64/sn/pci/pci_dma.c
@@ -10,7 +10,7 @@
  */
 
 #include <linux/module.h>
-#include <linux/dma-attrs.h>
+#include <linux/dma-mapping.h>
 #include <asm/dma.h>
 #include <asm/sn/intr.h>
 #include <asm/sn/pcibus_provider_defs.h>
@@ -31,7 +31,7 @@
  * this function.  Of course, SN only supports devices that have 32 or more
  * address bits when using the PMU.
  */
-int sn_dma_supported(struct device *dev, u64 mask)
+static int sn_dma_supported(struct device *dev, u64 mask)
 {
 	BUG_ON(dev->bus != &pci_bus_type);
 
@@ -39,7 +39,6 @@ int sn_dma_supported(struct device *dev, u64 mask)
 		return 0;
 	return 1;
 }
-EXPORT_SYMBOL(sn_dma_supported);
 
 /**
  * sn_dma_set_mask - set the DMA mask
@@ -75,8 +74,8 @@ EXPORT_SYMBOL(sn_dma_set_mask);
  * queue for a SCSI controller).  See Documentation/DMA-API.txt for
  * more information.
  */
-void *sn_dma_alloc_coherent(struct device *dev, size_t size,
-			    dma_addr_t * dma_handle, gfp_t flags)
+static void *sn_dma_alloc_coherent(struct device *dev, size_t size,
+				   dma_addr_t * dma_handle, gfp_t flags)
 {
 	void *cpuaddr;
 	unsigned long phys_addr;
@@ -124,7 +123,6 @@ void *sn_dma_alloc_coherent(struct device *dev, size_t size,
 
 	return cpuaddr;
 }
-EXPORT_SYMBOL(sn_dma_alloc_coherent);
 
 /**
  * sn_pci_free_coherent - free memory associated with coherent DMAable region
@@ -136,8 +134,8 @@ EXPORT_SYMBOL(sn_dma_alloc_coherent);
  * Frees the memory allocated by dma_alloc_coherent(), potentially unmapping
  * any associated IOMMU mappings.
  */
-void sn_dma_free_coherent(struct device *dev, size_t size, void *cpu_addr,
-			  dma_addr_t dma_handle)
+static void sn_dma_free_coherent(struct device *dev, size_t size, void *cpu_addr,
+				 dma_addr_t dma_handle)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct sn_pcibus_provider *provider = SN_PCIDEV_BUSPROVIDER(pdev);
@@ -147,7 +145,6 @@ void sn_dma_free_coherent(struct device *dev, size_t size, void *cpu_addr,
 	provider->dma_unmap(pdev, dma_handle, 0);
 	free_pages((unsigned long)cpu_addr, get_order(size));
 }
-EXPORT_SYMBOL(sn_dma_free_coherent);
 
 /**
  * sn_dma_map_single_attrs - map a single page for DMA
@@ -173,10 +170,12 @@ EXPORT_SYMBOL(sn_dma_free_coherent);
  * TODO: simplify our interface;
  *       figure out how to save dmamap handle so can use two step.
  */
-dma_addr_t sn_dma_map_single_attrs(struct device *dev, void *cpu_addr,
-				   size_t size, int direction,
-				   struct dma_attrs *attrs)
+static dma_addr_t sn_dma_map_page(struct device *dev, struct page *page,
+				  unsigned long offset, size_t size,
+				  enum dma_data_direction dir,
+				  struct dma_attrs *attrs)
 {
+	void *cpu_addr = page_address(page) + offset;
 	dma_addr_t dma_addr;
 	unsigned long phys_addr;
 	struct pci_dev *pdev = to_pci_dev(dev);
@@ -201,7 +200,6 @@ dma_addr_t sn_dma_map_single_attrs(struct device *dev, void *cpu_addr,
 	}
 	return dma_addr;
 }
-EXPORT_SYMBOL(sn_dma_map_single_attrs);
 
 /**
  * sn_dma_unmap_single_attrs - unamp a DMA mapped page
@@ -215,21 +213,20 @@ EXPORT_SYMBOL(sn_dma_map_single_attrs);
  * by @dma_handle into the coherence domain.  On SN, we're always cache
  * coherent, so we just need to free any ATEs associated with this mapping.
  */
-void sn_dma_unmap_single_attrs(struct device *dev, dma_addr_t dma_addr,
-			       size_t size, int direction,
-			       struct dma_attrs *attrs)
+static void sn_dma_unmap_page(struct device *dev, dma_addr_t dma_addr,
+			      size_t size, enum dma_data_direction dir,
+			      struct dma_attrs *attrs)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct sn_pcibus_provider *provider = SN_PCIDEV_BUSPROVIDER(pdev);
 
 	BUG_ON(dev->bus != &pci_bus_type);
 
-	provider->dma_unmap(pdev, dma_addr, direction);
+	provider->dma_unmap(pdev, dma_addr, dir);
 }
-EXPORT_SYMBOL(sn_dma_unmap_single_attrs);
 
 /**
- * sn_dma_unmap_sg_attrs - unmap a DMA scatterlist
+ * sn_dma_unmap_sg - unmap a DMA scatterlist
  * @dev: device to unmap
  * @sg: scatterlist to unmap
  * @nhwentries: number of scatterlist entries
@@ -238,9 +235,9 @@ EXPORT_SYMBOL(sn_dma_unmap_single_attrs);
  *
  * Unmap a set of streaming mode DMA translations.
  */
-void sn_dma_unmap_sg_attrs(struct device *dev, struct scatterlist *sgl,
-			   int nhwentries, int direction,
-			   struct dma_attrs *attrs)
+static void sn_dma_unmap_sg(struct device *dev, struct scatterlist *sgl,
+			    int nhwentries, enum dma_data_direction dir,
+			    struct dma_attrs *attrs)
 {
 	int i;
 	struct pci_dev *pdev = to_pci_dev(dev);
@@ -250,15 +247,14 @@ void sn_dma_unmap_sg_attrs(struct device *dev, struct scatterlist *sgl,
 	BUG_ON(dev->bus != &pci_bus_type);
 
 	for_each_sg(sgl, sg, nhwentries, i) {
-		provider->dma_unmap(pdev, sg->dma_address, direction);
+		provider->dma_unmap(pdev, sg->dma_address, dir);
 		sg->dma_address = (dma_addr_t) NULL;
 		sg->dma_length = 0;
 	}
 }
-EXPORT_SYMBOL(sn_dma_unmap_sg_attrs);
 
 /**
- * sn_dma_map_sg_attrs - map a scatterlist for DMA
+ * sn_dma_map_sg - map a scatterlist for DMA
  * @dev: device to map for
  * @sg: scatterlist to map
  * @nhwentries: number of entries
@@ -272,8 +268,9 @@ EXPORT_SYMBOL(sn_dma_unmap_sg_attrs);
  *
  * Maps each entry of @sg for DMA.
  */
-int sn_dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
-			int nhwentries, int direction, struct dma_attrs *attrs)
+static int sn_dma_map_sg(struct device *dev, struct scatterlist *sgl,
+			 int nhwentries, enum dma_data_direction dir,
+			 struct dma_attrs *attrs)
 {
 	unsigned long phys_addr;
 	struct scatterlist *saved_sg = sgl, *sg;
@@ -310,8 +307,7 @@ int sn_dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
 			 * Free any successfully allocated entries.
 			 */
 			if (i > 0)
-				sn_dma_unmap_sg_attrs(dev, saved_sg, i,
-						      direction, attrs);
+				sn_dma_unmap_sg(dev, saved_sg, i, dir, attrs);
 			return 0;
 		}
 
@@ -320,41 +316,36 @@ int sn_dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
 
 	return nhwentries;
 }
-EXPORT_SYMBOL(sn_dma_map_sg_attrs);
 
-void sn_dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle,
-				size_t size, int direction)
+static void sn_dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle,
+				       size_t size, enum dma_data_direction dir)
 {
 	BUG_ON(dev->bus != &pci_bus_type);
 }
-EXPORT_SYMBOL(sn_dma_sync_single_for_cpu);
 
-void sn_dma_sync_single_for_device(struct device *dev, dma_addr_t dma_handle,
-				   size_t size, int direction)
+static void sn_dma_sync_single_for_device(struct device *dev, dma_addr_t dma_handle,
+					  size_t size,
+					  enum dma_data_direction dir)
 {
 	BUG_ON(dev->bus != &pci_bus_type);
 }
-EXPORT_SYMBOL(sn_dma_sync_single_for_device);
 
-void sn_dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
-			    int nelems, int direction)
+static void sn_dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
+				   int nelems, enum dma_data_direction dir)
 {
 	BUG_ON(dev->bus != &pci_bus_type);
 }
-EXPORT_SYMBOL(sn_dma_sync_sg_for_cpu);
 
-void sn_dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
-			       int nelems, int direction)
+static void sn_dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
+				      int nelems, enum dma_data_direction dir)
 {
 	BUG_ON(dev->bus != &pci_bus_type);
 }
-EXPORT_SYMBOL(sn_dma_sync_sg_for_device);
 
-int sn_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
+static int sn_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
 {
 	return 0;
 }
-EXPORT_SYMBOL(sn_dma_mapping_error);
 
 u64 sn_dma_get_required_mask(struct device *dev)
 {
@@ -471,3 +462,23 @@ int sn_pci_legacy_write(struct pci_bus *bus, u16 port, u32 val, u8 size)
  out:
 	return ret;
 }
+
+static struct dma_map_ops sn_dma_ops = {
+	.alloc_coherent		= sn_dma_alloc_coherent,
+	.free_coherent		= sn_dma_free_coherent,
+	.map_page		= sn_dma_map_page,
+	.unmap_page		= sn_dma_unmap_page,
+	.map_sg			= sn_dma_map_sg,
+	.unmap_sg		= sn_dma_unmap_sg,
+	.sync_single_for_cpu 	= sn_dma_sync_single_for_cpu,
+	.sync_sg_for_cpu	= sn_dma_sync_sg_for_cpu,
+	.sync_single_for_device = sn_dma_sync_single_for_device,
+	.sync_sg_for_device	= sn_dma_sync_sg_for_device,
+	.mapping_error		= sn_dma_mapping_error,
+	.dma_supported		= sn_dma_supported,
+};
+
+void sn_dma_init(void)
+{
+	dma_ops = &sn_dma_ops;
+}
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 34bc3a89228b..45161b816313 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -40,6 +40,7 @@ config X86
 	select HAVE_GENERIC_DMA_COHERENT if X86_32
 	select HAVE_EFFICIENT_UNALIGNED_ACCESS
 	select USER_STACKTRACE_SUPPORT
+	select HAVE_DMA_API_DEBUG
 	select HAVE_KERNEL_GZIP
 	select HAVE_KERNEL_BZIP2
 	select HAVE_KERNEL_LZMA
diff --git a/arch/x86/include/asm/device.h b/arch/x86/include/asm/device.h
index 3c034f48fdb0..4994a20acbcb 100644
--- a/arch/x86/include/asm/device.h
+++ b/arch/x86/include/asm/device.h
@@ -6,7 +6,7 @@ struct dev_archdata {
 	void	*acpi_handle;
 #endif
 #ifdef CONFIG_X86_64
-struct dma_mapping_ops *dma_ops;
+struct dma_map_ops *dma_ops;
 #endif
 #ifdef CONFIG_DMAR
 	void *iommu; /* hook for IOMMU specific extension */
diff --git a/arch/x86/include/asm/dma-mapping.h b/arch/x86/include/asm/dma-mapping.h
index 132a134d12f2..cea7b74963e9 100644
--- a/arch/x86/include/asm/dma-mapping.h
+++ b/arch/x86/include/asm/dma-mapping.h
@@ -7,6 +7,8 @@
  */
 
 #include <linux/scatterlist.h>
+#include <linux/dma-debug.h>
+#include <linux/dma-attrs.h>
 #include <asm/io.h>
 #include <asm/swiotlb.h>
 #include <asm-generic/dma-coherent.h>
@@ -16,47 +18,9 @@ extern int iommu_merge;
 extern struct device x86_dma_fallback_dev;
 extern int panic_on_overflow;
 
-struct dma_mapping_ops {
-	int             (*mapping_error)(struct device *dev,
-					 dma_addr_t dma_addr);
-	void*           (*alloc_coherent)(struct device *dev, size_t size,
-				dma_addr_t *dma_handle, gfp_t gfp);
-	void            (*free_coherent)(struct device *dev, size_t size,
-				void *vaddr, dma_addr_t dma_handle);
-	dma_addr_t      (*map_single)(struct device *hwdev, phys_addr_t ptr,
-				size_t size, int direction);
-	void            (*unmap_single)(struct device *dev, dma_addr_t addr,
-				size_t size, int direction);
-	void            (*sync_single_for_cpu)(struct device *hwdev,
-				dma_addr_t dma_handle, size_t size,
-				int direction);
-	void            (*sync_single_for_device)(struct device *hwdev,
-				dma_addr_t dma_handle, size_t size,
-				int direction);
-	void            (*sync_single_range_for_cpu)(struct device *hwdev,
-				dma_addr_t dma_handle, unsigned long offset,
-				size_t size, int direction);
-	void            (*sync_single_range_for_device)(struct device *hwdev,
-				dma_addr_t dma_handle, unsigned long offset,
-				size_t size, int direction);
-	void            (*sync_sg_for_cpu)(struct device *hwdev,
-				struct scatterlist *sg, int nelems,
-				int direction);
-	void            (*sync_sg_for_device)(struct device *hwdev,
-				struct scatterlist *sg, int nelems,
-				int direction);
-	int             (*map_sg)(struct device *hwdev, struct scatterlist *sg,
-				int nents, int direction);
-	void            (*unmap_sg)(struct device *hwdev,
-				struct scatterlist *sg, int nents,
-				int direction);
-	int             (*dma_supported)(struct device *hwdev, u64 mask);
-	int		is_phys;
-};
-
-extern struct dma_mapping_ops *dma_ops;
-
-static inline struct dma_mapping_ops *get_dma_ops(struct device *dev)
+extern struct dma_map_ops *dma_ops;
+
+static inline struct dma_map_ops *get_dma_ops(struct device *dev)
 {
 #ifdef CONFIG_X86_32
 	return dma_ops;
@@ -71,7 +35,7 @@ static inline struct dma_mapping_ops *get_dma_ops(struct device *dev)
 /* Make sure we keep the same behaviour */
 static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(dev);
+	struct dma_map_ops *ops = get_dma_ops(dev);
 	if (ops->mapping_error)
 		return ops->mapping_error(dev, dma_addr);
 
@@ -90,137 +54,167 @@ extern void *dma_generic_alloc_coherent(struct device *dev, size_t size,
 
 static inline dma_addr_t
 dma_map_single(struct device *hwdev, void *ptr, size_t size,
-	       int direction)
+	       enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
-
-	BUG_ON(!valid_dma_direction(direction));
-	return ops->map_single(hwdev, virt_to_phys(ptr), size, direction);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
+	dma_addr_t addr;
+
+	BUG_ON(!valid_dma_direction(dir));
+	addr = ops->map_page(hwdev, virt_to_page(ptr),
+			     (unsigned long)ptr & ~PAGE_MASK, size,
+			     dir, NULL);
+	debug_dma_map_page(hwdev, virt_to_page(ptr),
+			   (unsigned long)ptr & ~PAGE_MASK, size,
+			   dir, addr, true);
+	return addr;
 }
 
 static inline void
 dma_unmap_single(struct device *dev, dma_addr_t addr, size_t size,
-		 int direction)
+		 enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(dev);
+	struct dma_map_ops *ops = get_dma_ops(dev);
 
-	BUG_ON(!valid_dma_direction(direction));
-	if (ops->unmap_single)
-		ops->unmap_single(dev, addr, size, direction);
+	BUG_ON(!valid_dma_direction(dir));
+	if (ops->unmap_page)
+		ops->unmap_page(dev, addr, size, dir, NULL);
+	debug_dma_unmap_page(dev, addr, size, dir, true);
 }
 
 static inline int
 dma_map_sg(struct device *hwdev, struct scatterlist *sg,
-	   int nents, int direction)
+	   int nents, enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
+	int ents;
+
+	BUG_ON(!valid_dma_direction(dir));
+	ents = ops->map_sg(hwdev, sg, nents, dir, NULL);
+	debug_dma_map_sg(hwdev, sg, nents, ents, dir);
 
-	BUG_ON(!valid_dma_direction(direction));
-	return ops->map_sg(hwdev, sg, nents, direction);
+	return ents;
 }
 
 static inline void
 dma_unmap_sg(struct device *hwdev, struct scatterlist *sg, int nents,
-	     int direction)
+	     enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
 
-	BUG_ON(!valid_dma_direction(direction));
+	BUG_ON(!valid_dma_direction(dir));
+	debug_dma_unmap_sg(hwdev, sg, nents, dir);
 	if (ops->unmap_sg)
-		ops->unmap_sg(hwdev, sg, nents, direction);
+		ops->unmap_sg(hwdev, sg, nents, dir, NULL);
 }
 
 static inline void
 dma_sync_single_for_cpu(struct device *hwdev, dma_addr_t dma_handle,
-			size_t size, int direction)
+			size_t size, enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
 
-	BUG_ON(!valid_dma_direction(direction));
+	BUG_ON(!valid_dma_direction(dir));
 	if (ops->sync_single_for_cpu)
-		ops->sync_single_for_cpu(hwdev, dma_handle, size, direction);
+		ops->sync_single_for_cpu(hwdev, dma_handle, size, dir);
+	debug_dma_sync_single_for_cpu(hwdev, dma_handle, size, dir);
 	flush_write_buffers();
 }
 
 static inline void
 dma_sync_single_for_device(struct device *hwdev, dma_addr_t dma_handle,
-			   size_t size, int direction)
+			   size_t size, enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
 
-	BUG_ON(!valid_dma_direction(direction));
+	BUG_ON(!valid_dma_direction(dir));
 	if (ops->sync_single_for_device)
-		ops->sync_single_for_device(hwdev, dma_handle, size, direction);
+		ops->sync_single_for_device(hwdev, dma_handle, size, dir);
+	debug_dma_sync_single_for_device(hwdev, dma_handle, size, dir);
 	flush_write_buffers();
 }
 
 static inline void
 dma_sync_single_range_for_cpu(struct device *hwdev, dma_addr_t dma_handle,
-			      unsigned long offset, size_t size, int direction)
+			      unsigned long offset, size_t size,
+			      enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
 
-	BUG_ON(!valid_dma_direction(direction));
+	BUG_ON(!valid_dma_direction(dir));
 	if (ops->sync_single_range_for_cpu)
 		ops->sync_single_range_for_cpu(hwdev, dma_handle, offset,
-					       size, direction);
+					       size, dir);
+	debug_dma_sync_single_range_for_cpu(hwdev, dma_handle,
+					    offset, size, dir);
 	flush_write_buffers();
 }
 
 static inline void
 dma_sync_single_range_for_device(struct device *hwdev, dma_addr_t dma_handle,
 				 unsigned long offset, size_t size,
-				 int direction)
+				 enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
 
-	BUG_ON(!valid_dma_direction(direction));
+	BUG_ON(!valid_dma_direction(dir));
 	if (ops->sync_single_range_for_device)
 		ops->sync_single_range_for_device(hwdev, dma_handle,
-						  offset, size, direction);
+						  offset, size, dir);
+	debug_dma_sync_single_range_for_device(hwdev, dma_handle,
+					       offset, size, dir);
 	flush_write_buffers();
 }
 
 static inline void
 dma_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg,
-		    int nelems, int direction)
+		    int nelems, enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
 
-	BUG_ON(!valid_dma_direction(direction));
+	BUG_ON(!valid_dma_direction(dir));
 	if (ops->sync_sg_for_cpu)
-		ops->sync_sg_for_cpu(hwdev, sg, nelems, direction);
+		ops->sync_sg_for_cpu(hwdev, sg, nelems, dir);
+	debug_dma_sync_sg_for_cpu(hwdev, sg, nelems, dir);
 	flush_write_buffers();
 }
 
 static inline void
 dma_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
-		       int nelems, int direction)
+		       int nelems, enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(hwdev);
+	struct dma_map_ops *ops = get_dma_ops(hwdev);
 
-	BUG_ON(!valid_dma_direction(direction));
+	BUG_ON(!valid_dma_direction(dir));
 	if (ops->sync_sg_for_device)
-		ops->sync_sg_for_device(hwdev, sg, nelems, direction);
+		ops->sync_sg_for_device(hwdev, sg, nelems, dir);
+	debug_dma_sync_sg_for_device(hwdev, sg, nelems, dir);
 
 	flush_write_buffers();
 }
 
 static inline dma_addr_t dma_map_page(struct device *dev, struct page *page,
 				      size_t offset, size_t size,
-				      int direction)
+				      enum dma_data_direction dir)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(dev);
+	struct dma_map_ops *ops = get_dma_ops(dev);
+	dma_addr_t addr;
 
-	BUG_ON(!valid_dma_direction(direction));
-	return ops->map_single(dev, page_to_phys(page) + offset,
-			       size, direction);
+	BUG_ON(!valid_dma_direction(dir));
+	addr = ops->map_page(dev, page, offset, size, dir, NULL);
+	debug_dma_map_page(dev, page, offset, size, dir, addr, false);
+
+	return addr;
 }
 
 static inline void dma_unmap_page(struct device *dev, dma_addr_t addr,
-				  size_t size, int direction)
+				  size_t size, enum dma_data_direction dir)
 {
-	dma_unmap_single(dev, addr, size, direction);
+	struct dma_map_ops *ops = get_dma_ops(dev);
+
+	BUG_ON(!valid_dma_direction(dir));
+	if (ops->unmap_page)
+		ops->unmap_page(dev, addr, size, dir, NULL);
+	debug_dma_unmap_page(dev, addr, size, dir, false);
 }
 
 static inline void
@@ -266,7 +260,7 @@ static inline void *
 dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
 		gfp_t gfp)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(dev);
+	struct dma_map_ops *ops = get_dma_ops(dev);
 	void *memory;
 
 	gfp &= ~(__GFP_DMA | __GFP_HIGHMEM | __GFP_DMA32);
@@ -285,20 +279,24 @@ dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
 	if (!ops->alloc_coherent)
 		return NULL;
 
-	return ops->alloc_coherent(dev, size, dma_handle,
-				   dma_alloc_coherent_gfp_flags(dev, gfp));
+	memory = ops->alloc_coherent(dev, size, dma_handle,
+				     dma_alloc_coherent_gfp_flags(dev, gfp));
+	debug_dma_alloc_coherent(dev, size, *dma_handle, memory);
+
+	return memory;
 }
 
 static inline void dma_free_coherent(struct device *dev, size_t size,
 				     void *vaddr, dma_addr_t bus)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(dev);
+	struct dma_map_ops *ops = get_dma_ops(dev);
 
 	WARN_ON(irqs_disabled());       /* for portability */
 
 	if (dma_release_from_coherent(dev, get_order(size), vaddr))
 		return;
 
+	debug_dma_free_coherent(dev, size, vaddr, bus);
 	if (ops->free_coherent)
 		ops->free_coherent(dev, size, vaddr, bus);
 }
diff --git a/arch/x86/include/asm/iommu.h b/arch/x86/include/asm/iommu.h
index a6ee9e6f530f..af326a2975b5 100644
--- a/arch/x86/include/asm/iommu.h
+++ b/arch/x86/include/asm/iommu.h
@@ -3,7 +3,7 @@
 
 extern void pci_iommu_shutdown(void);
 extern void no_iommu_init(void);
-extern struct dma_mapping_ops nommu_dma_ops;
+extern struct dma_map_ops nommu_dma_ops;
 extern int force_iommu, no_iommu;
 extern int iommu_detected;
 
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index 6e9c1f320acf..c611ad64137f 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -105,7 +105,7 @@ obj-$(CONFIG_MICROCODE)			+= microcode.o
 
 obj-$(CONFIG_X86_CHECK_BIOS_CORRUPTION) += check.o
 
-obj-$(CONFIG_SWIOTLB)			+= pci-swiotlb_64.o # NB rename without _64
+obj-$(CONFIG_SWIOTLB)			+= pci-swiotlb.o
 
 ###
 # 64 bit specific files
diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c
index 5113c080f0c4..c5962fe3796f 100644
--- a/arch/x86/kernel/amd_iommu.c
+++ b/arch/x86/kernel/amd_iommu.c
@@ -22,10 +22,9 @@
 #include <linux/bitops.h>
 #include <linux/debugfs.h>
 #include <linux/scatterlist.h>
+#include <linux/dma-mapping.h>
 #include <linux/iommu-helper.h>
-#ifdef CONFIG_IOMMU_API
 #include <linux/iommu.h>
-#endif
 #include <asm/proto.h>
 #include <asm/iommu.h>
 #include <asm/gart.h>
@@ -1297,8 +1296,10 @@ static void __unmap_single(struct amd_iommu *iommu,
 /*
  * The exported map_single function for dma_ops.
  */
-static dma_addr_t map_single(struct device *dev, phys_addr_t paddr,
-			     size_t size, int dir)
+static dma_addr_t map_page(struct device *dev, struct page *page,
+			   unsigned long offset, size_t size,
+			   enum dma_data_direction dir,
+			   struct dma_attrs *attrs)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;
@@ -1306,6 +1307,7 @@ static dma_addr_t map_single(struct device *dev, phys_addr_t paddr,
 	u16 devid;
 	dma_addr_t addr;
 	u64 dma_mask;
+	phys_addr_t paddr = page_to_phys(page) + offset;
 
 	INC_STATS_COUNTER(cnt_map_single);
 
@@ -1340,8 +1342,8 @@ out:
 /*
  * The exported unmap_single function for dma_ops.
  */
-static void unmap_single(struct device *dev, dma_addr_t dma_addr,
-			 size_t size, int dir)
+static void unmap_page(struct device *dev, dma_addr_t dma_addr, size_t size,
+		       enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;
@@ -1390,7 +1392,8 @@ static int map_sg_no_iommu(struct device *dev, struct scatterlist *sglist,
  * lists).
  */
 static int map_sg(struct device *dev, struct scatterlist *sglist,
-		  int nelems, int dir)
+		  int nelems, enum dma_data_direction dir,
+		  struct dma_attrs *attrs)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;
@@ -1457,7 +1460,8 @@ unmap:
  * lists).
  */
 static void unmap_sg(struct device *dev, struct scatterlist *sglist,
-		     int nelems, int dir)
+		     int nelems, enum dma_data_direction dir,
+		     struct dma_attrs *attrs)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;
@@ -1644,11 +1648,11 @@ static void prealloc_protection_domains(void)
 	}
 }
 
-static struct dma_mapping_ops amd_iommu_dma_ops = {
+static struct dma_map_ops amd_iommu_dma_ops = {
 	.alloc_coherent = alloc_coherent,
 	.free_coherent = free_coherent,
-	.map_single = map_single,
-	.unmap_single = unmap_single,
+	.map_page = map_page,
+	.unmap_page = unmap_page,
 	.map_sg = map_sg,
 	.unmap_sg = unmap_sg,
 	.dma_supported = amd_iommu_dma_supported,
diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c
index d28bbdc35e4e..755c21e906f3 100644
--- a/arch/x86/kernel/pci-calgary_64.c
+++ b/arch/x86/kernel/pci-calgary_64.c
@@ -380,8 +380,9 @@ static inline struct iommu_table *find_iommu_table(struct device *dev)
 	return tbl;
 }
 
-static void calgary_unmap_sg(struct device *dev,
-	struct scatterlist *sglist, int nelems, int direction)
+static void calgary_unmap_sg(struct device *dev, struct scatterlist *sglist,
+			     int nelems,enum dma_data_direction dir,
+			     struct dma_attrs *attrs)
 {
 	struct iommu_table *tbl = find_iommu_table(dev);
 	struct scatterlist *s;
@@ -404,7 +405,8 @@ static void calgary_unmap_sg(struct device *dev,
 }
 
 static int calgary_map_sg(struct device *dev, struct scatterlist *sg,
-	int nelems, int direction)
+			  int nelems, enum dma_data_direction dir,
+			  struct dma_attrs *attrs)
 {
 	struct iommu_table *tbl = find_iommu_table(dev);
 	struct scatterlist *s;
@@ -429,15 +431,14 @@ static int calgary_map_sg(struct device *dev, struct scatterlist *sg,
 		s->dma_address = (entry << PAGE_SHIFT) | s->offset;
 
 		/* insert into HW table */
-		tce_build(tbl, entry, npages, vaddr & PAGE_MASK,
-			  direction);
+		tce_build(tbl, entry, npages, vaddr & PAGE_MASK, dir);
 
 		s->dma_length = s->length;
 	}
 
 	return nelems;
 error:
-	calgary_unmap_sg(dev, sg, nelems, direction);
+	calgary_unmap_sg(dev, sg, nelems, dir, NULL);
 	for_each_sg(sg, s, nelems, i) {
 		sg->dma_address = bad_dma_address;
 		sg->dma_length = 0;
@@ -445,10 +446,12 @@ error:
 	return 0;
 }
 
-static dma_addr_t calgary_map_single(struct device *dev, phys_addr_t paddr,
-	size_t size, int direction)
+static dma_addr_t calgary_map_page(struct device *dev, struct page *page,
+				   unsigned long offset, size_t size,
+				   enum dma_data_direction dir,
+				   struct dma_attrs *attrs)
 {
-	void *vaddr = phys_to_virt(paddr);
+	void *vaddr = page_address(page) + offset;
 	unsigned long uaddr;
 	unsigned int npages;
 	struct iommu_table *tbl = find_iommu_table(dev);
@@ -456,17 +459,18 @@ static dma_addr_t calgary_map_single(struct device *dev, phys_addr_t paddr,
 	uaddr = (unsigned long)vaddr;
 	npages = iommu_num_pages(uaddr, size, PAGE_SIZE);
 
-	return iommu_alloc(dev, tbl, vaddr, npages, direction);
+	return iommu_alloc(dev, tbl, vaddr, npages, dir);
 }
 
-static void calgary_unmap_single(struct device *dev, dma_addr_t dma_handle,
-	size_t size, int direction)
+static void calgary_unmap_page(struct device *dev, dma_addr_t dma_addr,
+			       size_t size, enum dma_data_direction dir,
+			       struct dma_attrs *attrs)
 {
 	struct iommu_table *tbl = find_iommu_table(dev);
 	unsigned int npages;
 
-	npages = iommu_num_pages(dma_handle, size, PAGE_SIZE);
-	iommu_free(tbl, dma_handle, npages);
+	npages = iommu_num_pages(dma_addr, size, PAGE_SIZE);
+	iommu_free(tbl, dma_addr, npages);
 }
 
 static void* calgary_alloc_coherent(struct device *dev, size_t size,
@@ -515,13 +519,13 @@ static void calgary_free_coherent(struct device *dev, size_t size,
 	free_pages((unsigned long)vaddr, get_order(size));
 }
 
-static struct dma_mapping_ops calgary_dma_ops = {
+static struct dma_map_ops calgary_dma_ops = {
 	.alloc_coherent = calgary_alloc_coherent,
 	.free_coherent = calgary_free_coherent,
-	.map_single = calgary_map_single,
-	.unmap_single = calgary_unmap_single,
 	.map_sg = calgary_map_sg,
 	.unmap_sg = calgary_unmap_sg,
+	.map_page = calgary_map_page,
+	.unmap_page = calgary_unmap_page,
 };
 
 static inline void __iomem * busno_to_bbar(unsigned char num)
diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c
index b25428533141..c7c4776ff630 100644
--- a/arch/x86/kernel/pci-dma.c
+++ b/arch/x86/kernel/pci-dma.c
@@ -1,4 +1,5 @@
 #include <linux/dma-mapping.h>
+#include <linux/dma-debug.h>
 #include <linux/dmar.h>
 #include <linux/bootmem.h>
 #include <linux/pci.h>
@@ -12,7 +13,7 @@
 
 static int forbid_dac __read_mostly;
 
-struct dma_mapping_ops *dma_ops;
+struct dma_map_ops *dma_ops;
 EXPORT_SYMBOL(dma_ops);
 
 static int iommu_sac_force __read_mostly;
@@ -44,6 +45,9 @@ struct device x86_dma_fallback_dev = {
 };
 EXPORT_SYMBOL(x86_dma_fallback_dev);
 
+/* Number of entries preallocated for DMA-API debugging */
+#define PREALLOC_DMA_DEBUG_ENTRIES       32768
+
 int dma_set_mask(struct device *dev, u64 mask)
 {
 	if (!dev->dma_mask || !dma_supported(dev, mask))
@@ -224,7 +228,7 @@ early_param("iommu", iommu_setup);
 
 int dma_supported(struct device *dev, u64 mask)
 {
-	struct dma_mapping_ops *ops = get_dma_ops(dev);
+	struct dma_map_ops *ops = get_dma_ops(dev);
 
 #ifdef CONFIG_PCI
 	if (mask > 0xffffffff && forbid_dac > 0) {
@@ -265,6 +269,12 @@ EXPORT_SYMBOL(dma_supported);
 
 static int __init pci_iommu_init(void)
 {
+	dma_debug_init(PREALLOC_DMA_DEBUG_ENTRIES);
+
+#ifdef CONFIG_PCI
+	dma_debug_add_bus(&pci_bus_type);
+#endif
+
 	calgary_iommu_init();
 
 	intel_iommu_init();
diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c
index d5768b1af080..b284b58c035c 100644
--- a/arch/x86/kernel/pci-gart_64.c
+++ b/arch/x86/kernel/pci-gart_64.c
@@ -255,10 +255,13 @@ static dma_addr_t dma_map_area(struct device *dev, dma_addr_t phys_mem,
 }
 
 /* Map a single area into the IOMMU */
-static dma_addr_t
-gart_map_single(struct device *dev, phys_addr_t paddr, size_t size, int dir)
+static dma_addr_t gart_map_page(struct device *dev, struct page *page,
+				unsigned long offset, size_t size,
+				enum dma_data_direction dir,
+				struct dma_attrs *attrs)
 {
 	unsigned long bus;
+	phys_addr_t paddr = page_to_phys(page) + offset;
 
 	if (!dev)
 		dev = &x86_dma_fallback_dev;
@@ -275,8 +278,9 @@ gart_map_single(struct device *dev, phys_addr_t paddr, size_t size, int dir)
 /*
  * Free a DMA mapping.
  */
-static void gart_unmap_single(struct device *dev, dma_addr_t dma_addr,
-			      size_t size, int direction)
+static void gart_unmap_page(struct device *dev, dma_addr_t dma_addr,
+			    size_t size, enum dma_data_direction dir,
+			    struct dma_attrs *attrs)
 {
 	unsigned long iommu_page;
 	int npages;
@@ -298,8 +302,8 @@ static void gart_unmap_single(struct device *dev, dma_addr_t dma_addr,
 /*
  * Wrapper for pci_unmap_single working with scatterlists.
  */
-static void
-gart_unmap_sg(struct device *dev, struct scatterlist *sg, int nents, int dir)
+static void gart_unmap_sg(struct device *dev, struct scatterlist *sg, int nents,
+			  enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	struct scatterlist *s;
 	int i;
@@ -307,7 +311,7 @@ gart_unmap_sg(struct device *dev, struct scatterlist *sg, int nents, int dir)
 	for_each_sg(sg, s, nents, i) {
 		if (!s->dma_length || !s->length)
 			break;
-		gart_unmap_single(dev, s->dma_address, s->dma_length, dir);
+		gart_unmap_page(dev, s->dma_address, s->dma_length, dir, NULL);
 	}
 }
 
@@ -329,7 +333,7 @@ static int dma_map_sg_nonforce(struct device *dev, struct scatterlist *sg,
 			addr = dma_map_area(dev, addr, s->length, dir, 0);
 			if (addr == bad_dma_address) {
 				if (i > 0)
-					gart_unmap_sg(dev, sg, i, dir);
+					gart_unmap_sg(dev, sg, i, dir, NULL);
 				nents = 0;
 				sg[0].dma_length = 0;
 				break;
@@ -400,8 +404,8 @@ dma_map_cont(struct device *dev, struct scatterlist *start, int nelems,
  * DMA map all entries in a scatterlist.
  * Merge chunks that have page aligned sizes into a continuous mapping.
  */
-static int
-gart_map_sg(struct device *dev, struct scatterlist *sg, int nents, int dir)
+static int gart_map_sg(struct device *dev, struct scatterlist *sg, int nents,
+		       enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	struct scatterlist *s, *ps, *start_sg, *sgmap;
 	int need = 0, nextneed, i, out, start;
@@ -468,7 +472,7 @@ gart_map_sg(struct device *dev, struct scatterlist *sg, int nents, int dir)
 
 error:
 	flush_gart();
-	gart_unmap_sg(dev, sg, out, dir);
+	gart_unmap_sg(dev, sg, out, dir, NULL);
 
 	/* When it was forced or merged try again in a dumb way */
 	if (force_iommu || iommu_merge) {
@@ -521,7 +525,7 @@ static void
 gart_free_coherent(struct device *dev, size_t size, void *vaddr,
 		   dma_addr_t dma_addr)
 {
-	gart_unmap_single(dev, dma_addr, size, DMA_BIDIRECTIONAL);
+	gart_unmap_page(dev, dma_addr, size, DMA_BIDIRECTIONAL, NULL);
 	free_pages((unsigned long)vaddr, get_order(size));
 }
 
@@ -707,11 +711,11 @@ static __init int init_k8_gatt(struct agp_kern_info *info)
 	return -1;
 }
 
-static struct dma_mapping_ops gart_dma_ops = {
-	.map_single			= gart_map_single,
-	.unmap_single			= gart_unmap_single,
+static struct dma_map_ops gart_dma_ops = {
 	.map_sg				= gart_map_sg,
 	.unmap_sg			= gart_unmap_sg,
+	.map_page			= gart_map_page,
+	.unmap_page			= gart_unmap_page,
 	.alloc_coherent			= gart_alloc_coherent,
 	.free_coherent			= gart_free_coherent,
 };
diff --git a/arch/x86/kernel/pci-nommu.c b/arch/x86/kernel/pci-nommu.c
index 8b02a3936d42..c6d703b39326 100644
--- a/arch/x86/kernel/pci-nommu.c
+++ b/arch/x86/kernel/pci-nommu.c
@@ -25,19 +25,19 @@ check_addr(char *name, struct device *hwdev, dma_addr_t bus, size_t size)
 	return 1;
 }
 
-static dma_addr_t
-nommu_map_single(struct device *hwdev, phys_addr_t paddr, size_t size,
-	       int direction)
+static dma_addr_t nommu_map_page(struct device *dev, struct page *page,
+				 unsigned long offset, size_t size,
+				 enum dma_data_direction dir,
+				 struct dma_attrs *attrs)
 {
-	dma_addr_t bus = paddr;
+	dma_addr_t bus = page_to_phys(page) + offset;
 	WARN_ON(size == 0);
-	if (!check_addr("map_single", hwdev, bus, size))
-				return bad_dma_address;
+	if (!check_addr("map_single", dev, bus, size))
+		return bad_dma_address;
 	flush_write_buffers();
 	return bus;
 }
 
-
 /* Map a set of buffers described by scatterlist in streaming
  * mode for DMA.  This is the scatter-gather version of the
  * above pci_map_single interface.  Here the scatter gather list
@@ -54,7 +54,8 @@ nommu_map_single(struct device *hwdev, phys_addr_t paddr, size_t size,
  * the same here.
  */
 static int nommu_map_sg(struct device *hwdev, struct scatterlist *sg,
-	       int nents, int direction)
+			int nents, enum dma_data_direction dir,
+			struct dma_attrs *attrs)
 {
 	struct scatterlist *s;
 	int i;
@@ -78,11 +79,11 @@ static void nommu_free_coherent(struct device *dev, size_t size, void *vaddr,
 	free_pages((unsigned long)vaddr, get_order(size));
 }
 
-struct dma_mapping_ops nommu_dma_ops = {
+struct dma_map_ops nommu_dma_ops = {
 	.alloc_coherent	= dma_generic_alloc_coherent,
 	.free_coherent	= nommu_free_coherent,
-	.map_single	= nommu_map_single,
 	.map_sg		= nommu_map_sg,
+	.map_page	= nommu_map_page,
 	.is_phys	= 1,
 };
 
diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb.c
index d59c91747665..34f12e9996ed 100644
--- a/arch/x86/kernel/pci-swiotlb_64.c
+++ b/arch/x86/kernel/pci-swiotlb.c
@@ -33,18 +33,11 @@ phys_addr_t swiotlb_bus_to_phys(dma_addr_t baddr)
 	return baddr;
 }
 
-int __weak swiotlb_arch_range_needs_mapping(void *ptr, size_t size)
+int __weak swiotlb_arch_range_needs_mapping(phys_addr_t paddr, size_t size)
 {
 	return 0;
 }
 
-static dma_addr_t
-swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size,
-			int direction)
-{
-	return swiotlb_map_single(hwdev, phys_to_virt(paddr), size, direction);
-}
-
 static void *x86_swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 					dma_addr_t *dma_handle, gfp_t flags)
 {
@@ -57,20 +50,20 @@ static void *x86_swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 	return swiotlb_alloc_coherent(hwdev, size, dma_handle, flags);
 }
 
-struct dma_mapping_ops swiotlb_dma_ops = {
+struct dma_map_ops swiotlb_dma_ops = {
 	.mapping_error = swiotlb_dma_mapping_error,
 	.alloc_coherent = x86_swiotlb_alloc_coherent,
 	.free_coherent = swiotlb_free_coherent,
-	.map_single = swiotlb_map_single_phys,
-	.unmap_single = swiotlb_unmap_single,
 	.sync_single_for_cpu = swiotlb_sync_single_for_cpu,
 	.sync_single_for_device = swiotlb_sync_single_for_device,
 	.sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
 	.sync_single_range_for_device = swiotlb_sync_single_range_for_device,
 	.sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
 	.sync_sg_for_device = swiotlb_sync_sg_for_device,
-	.map_sg = swiotlb_map_sg,
-	.unmap_sg = swiotlb_unmap_sg,
+	.map_sg = swiotlb_map_sg_attrs,
+	.unmap_sg = swiotlb_unmap_sg_attrs,
+	.map_page = swiotlb_map_page,
+	.unmap_page = swiotlb_unmap_page,
 	.dma_supported = NULL,
 };