swiotlb/x86: lay groundwork for xen dom0 use of swiotlb

@@ -7,8 +7,6 @@
 extern int force_iommu, no_iommu;
 extern int iommu_detected;
 
-extern unsigned long iommu_nr_pages(unsigned long addr, unsigned long len);
-
 /* 10 seconds */
 #define DMAR_OPERATION_TIMEOUT ((cycles_t) tsc_khz*10*1000)
 

@@ -121,13 +121,6 @@
 	pci_swiotlb_init();
 }
 
-unsigned long iommu_nr_pages(unsigned long addr, unsigned long len)
-{
-	unsigned long size = roundup((addr & ~PAGE_MASK) + len, PAGE_SIZE);
-
-	return size >> PAGE_SHIFT;
-}
-EXPORT_SYMBOL(iommu_nr_pages);
 #endif
 
 void *dma_generic_alloc_coherent(struct device *dev, size_t size,

@@ -10,6 +10,9 @@
 extern void
 swiotlb_init(void);
 
+extern void *swiotlb_alloc_boot(size_t bytes, unsigned long nslabs);
+extern void *swiotlb_alloc(unsigned order, unsigned long nslabs);
+
 extern void
 *swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 			dma_addr_t *dma_handle, gfp_t flags);

@@ -21,6 +21,7 @@
 #include <linux/mm.h>
 #include <linux/module.h>
 #include <linux/spinlock.h>
+#include <linux/swiotlb.h>
 #include <linux/string.h>
 #include <linux/types.h>
 #include <linux/ctype.h>

@@ -126,6 +127,16 @@
 __setup("swiotlb=", setup_io_tlb_npages);
 /* make io_tlb_overflow tunable too? */
 
+void * __weak swiotlb_alloc_boot(size_t size, unsigned long nslabs)
+{
+	return alloc_bootmem_low_pages(size);
+}
+
+void * __weak swiotlb_alloc(unsigned order, unsigned long nslabs)
+{
+	return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order);
+}
+
 /*
  * Statically reserve bounce buffer space and initialize bounce buffer data
  * structures for the software IO TLB used to implement the DMA API.

@@ -145,7 +156,7 @@
 	/*
 	 * Get IO TLB memory from the low pages
 	 */
-	io_tlb_start = alloc_bootmem_low_pages(bytes);
+	io_tlb_start = swiotlb_alloc_boot(bytes, io_tlb_nslabs);
 	if (!io_tlb_start)
 		panic("Cannot allocate SWIOTLB buffer");
 	io_tlb_end = io_tlb_start + bytes;

@@ -202,8 +213,7 @@
 	bytes = io_tlb_nslabs << IO_TLB_SHIFT;
 
 	while ((SLABS_PER_PAGE << order) > IO_TLB_MIN_SLABS) {
-		io_tlb_start = (char *)__get_free_pages(GFP_DMA | __GFP_NOWARN,
-		                                        order);
+		io_tlb_start = swiotlb_alloc(order, io_tlb_nslabs);
 		if (io_tlb_start)
 			break;
 		order--;

@@ -7,6 +7,20 @@
 struct dma_attrs;
 struct scatterlist;
 
+/*
+ * Maximum allowable number of contiguous slabs to map,
+ * must be a power of 2.  What is the appropriate value ?
+ * The complexity of {map,unmap}_single is linearly dependent on this value.
+ */
+#define IO_TLB_SEGSIZE	128
+
+
+/*
+ * log of the size of each IO TLB slab.  The number of slabs is command line
+ * controllable.
+ */
+#define IO_TLB_SHIFT 11
+
 extern void
 swiotlb_init(void);
 

@@ -23,6 +23,7 @@
 #include <linux/spinlock.h>
 #include <linux/swiotlb.h>
 #include <linux/string.h>
+#include <linux/swiotlb.h>
 #include <linux/types.h>
 #include <linux/ctype.h>
 

@@ -40,19 +41,6 @@
 #define SG_ENT_VIRT_ADDRESS(sg)	(sg_virt((sg)))
 #define SG_ENT_PHYS_ADDRESS(sg)	virt_to_bus(SG_ENT_VIRT_ADDRESS(sg))
 
-/*
- * Maximum allowable number of contiguous slabs to map,
- * must be a power of 2.  What is the appropriate value ?
- * The complexity of {map,unmap}_single is linearly dependent on this value.
- */
-#define IO_TLB_SEGSIZE	128
-
-/*
- * log of the size of each IO TLB slab.  The number of slabs is command line
- * controllable.
- */
-#define IO_TLB_SHIFT 11
-
 #define SLABS_PER_PAGE (1 << (PAGE_SHIFT - IO_TLB_SHIFT))
 
 /*

@@ -465,13 +465,9 @@
 	dma_addr_t dev_addr;
 	void *ret;
 	int order = get_order(size);
-	u64 dma_mask = DMA_32BIT_MASK;
-
-	if (hwdev && hwdev->coherent_dma_mask)
-		dma_mask = hwdev->coherent_dma_mask;
 
 	ret = (void *)__get_free_pages(flags, order);
-	if (ret && !is_buffer_dma_capable(dma_mask, virt_to_bus(ret), size)) {
+	if (ret && !address_needs_mapping(hwdev, virt_to_bus(ret), size)) {
 		/*
 		 * The allocated memory isn't reachable by the device.
 		 * Fall back on swiotlb_map_single().

@@ -495,9 +491,9 @@
 	dev_addr = virt_to_bus(ret);
 
 	/* Confirm address can be DMA'd by device */
-	if (!is_buffer_dma_capable(dma_mask, dev_addr, size)) {
+	if (!address_needs_mapping(hwdev, dev_addr, size)) {
 		printk("hwdev DMA mask = 0x%016Lx, dev_addr = 0x%016Lx\n",
-		       (unsigned long long)dma_mask,
+		       (unsigned long long)dma_get_mask(hwdev),
 		       (unsigned long long)dev_addr);
 
 		/* DMA_TO_DEVICE to avoid memcpy in unmap_single */

@@ -301,6 +301,10 @@
 	start_dma_addr = virt_to_bus(io_tlb_start) & mask;
 
 	offset_slots = ALIGN(start_dma_addr, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT;
+
+	/*
+         * Carefully handle integer overflow which can occur when mask == ~0UL.
+         */
 	max_slots = mask + 1
 		    ? ALIGN(mask + 1, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT
 		    : 1UL << (BITS_PER_LONG - IO_TLB_SHIFT);

@@ -27,6 +27,9 @@
 extern void *swiotlb_alloc_boot(size_t bytes, unsigned long nslabs);
 extern void *swiotlb_alloc(unsigned order, unsigned long nslabs);
 
+extern dma_addr_t swiotlb_phys_to_bus(phys_addr_t address);
+extern phys_addr_t swiotlb_bus_to_phys(dma_addr_t address);
+
 extern void
 *swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 			dma_addr_t *dma_handle, gfp_t flags);

154 lines not on this page · raw message

@@ -30,6 +30,8 @@
 extern dma_addr_t swiotlb_phys_to_bus(phys_addr_t address);
 extern phys_addr_t swiotlb_bus_to_phys(dma_addr_t address);
 
+extern int swiotlb_arch_range_needs_mapping(void *ptr, size_t size);
+
 extern void
 *swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 			dma_addr_t *dma_handle, gfp_t flags);

@@ -145,6 +145,11 @@
 	return phys_to_virt(swiotlb_bus_to_phys(address));
 }
 
+int __weak swiotlb_arch_range_needs_mapping(void *ptr, size_t size)
+{
+	return 0;
+}
+
 /*
  * Statically reserve bounce buffer space and initialize bounce buffer data
  * structures for the software IO TLB used to implement the DMA API.

@@ -297,6 +302,11 @@
 	return !is_buffer_dma_capable(dma_get_mask(hwdev), addr, size);
 }
 
+static inline int range_needs_mapping(void *ptr, size_t size)
+{
+	return swiotlb_force || swiotlb_arch_range_needs_mapping(ptr, size);
+}
+
 static int is_swiotlb_buffer(char *addr)
 {
 	return addr >= io_tlb_start && addr < io_tlb_end;

@@ -581,7 +591,8 @@
 	 * we can safely return the device addr and not worry about bounce
 	 * buffering it.
 	 */
-	if (!address_needs_mapping(hwdev, dev_addr, size) && !swiotlb_force)
+	if (!address_needs_mapping(hwdev, dev_addr, size) &&
+	    !range_needs_mapping(ptr, size))
 		return dev_addr;
 
 	/*

@@ -741,7 +752,7 @@
 	for_each_sg(sgl, sg, nelems, i) {
 		addr = SG_ENT_VIRT_ADDRESS(sg);
 		dev_addr = swiotlb_virt_to_bus(addr);
-		if (swiotlb_force ||
+		if (range_needs_mapping(sg_virt(sg), sg->length) ||
 		    address_needs_mapping(hwdev, dev_addr, sg->length)) {
 			void *map = map_single(hwdev, addr, sg->length, dir);
 			if (!map) {

@@ -312,6 +312,15 @@
 	return addr >= io_tlb_start && addr < io_tlb_end;
 }
 
+static void
+__sync_single(char *buffer, char *dma_addr, size_t size, int dir)
+{
+	if (dir == DMA_TO_DEVICE)
+		memcpy(dma_addr, buffer, size);
+	else
+		memcpy(buffer, dma_addr, size);
+}
+
 /*
  * Allocates bounce buffer and returns its kernel virtual address.
  */

@@ -413,7 +422,7 @@
 	for (i = 0; i < nslots; i++)
 		io_tlb_orig_addr[index+i] = buffer + (i << IO_TLB_SHIFT);
 	if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)
-		memcpy(dma_addr, buffer, size);
+		__sync_single(buffer, dma_addr, size, DMA_TO_DEVICE);
 
 	return dma_addr;
 }

@@ -437,7 +446,7 @@
 		 * bounce... copy the data back into the original buffer * and
 		 * delete the bounce buffer.
 		 */
-		memcpy(buffer, dma_addr, size);
+		__sync_single(buffer, dma_addr, size, DMA_FROM_DEVICE);
 
 	/*
 	 * Return the buffer to the free list by setting the corresponding

@@ -477,13 +486,13 @@
 	switch (target) {
 	case SYNC_FOR_CPU:
 		if (likely(dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL))
-			memcpy(buffer, dma_addr, size);
+			__sync_single(buffer, dma_addr, size, DMA_FROM_DEVICE);
 		else
 			BUG_ON(dir != DMA_TO_DEVICE);
 		break;
 	case SYNC_FOR_DEVICE:
 		if (likely(dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL))
-			memcpy(dma_addr, buffer, size);
+			__sync_single(buffer, dma_addr, size, DMA_TO_DEVICE);
 		else
 			BUG_ON(dir != DMA_FROM_DEVICE);
 		break;

@@ -26,6 +26,7 @@
 #include <linux/swiotlb.h>
 #include <linux/types.h>
 #include <linux/ctype.h>
+#include <linux/highmem.h>
 
 #include <asm/io.h>
 #include <asm/dma.h>

@@ -38,9 +39,6 @@
 #define OFFSET(val,align) ((unsigned long)	\
 	                   ( (val) & ( (align) - 1)))
 
-#define SG_ENT_VIRT_ADDRESS(sg)	(sg_virt((sg)))
-#define SG_ENT_PHYS_ADDRESS(sg)	virt_to_bus(SG_ENT_VIRT_ADDRESS(sg))
-
 #define SLABS_PER_PAGE (1 << (PAGE_SHIFT - IO_TLB_SHIFT))
 
 /*

@@ -91,7 +89,10 @@
  * We need to save away the original address corresponding to a mapped entry
  * for the sync operations.
  */
-static unsigned char **io_tlb_orig_addr;
+static struct swiotlb_phys_addr {
+	struct page *page;
+	unsigned int offset;
+} *io_tlb_orig_addr;
 
 /*
  * Protect the above data structures in the map and unmap calls

@@ -150,6 +151,11 @@
 	return 0;
 }
 
+static dma_addr_t swiotlb_sg_to_bus(struct scatterlist *sg)
+{
+	return swiotlb_phys_to_bus(page_to_phys(sg_page(sg)) + sg->offset);
+}
+
 /*
  * Statically reserve bounce buffer space and initialize bounce buffer data
  * structures for the software IO TLB used to implement the DMA API.

@@ -183,7 +189,7 @@
 	for (i = 0; i < io_tlb_nslabs; i++)
  		io_tlb_list[i] = IO_TLB_SEGSIZE - OFFSET(i, IO_TLB_SEGSIZE);
 	io_tlb_index = 0;
-	io_tlb_orig_addr = alloc_bootmem(io_tlb_nslabs * sizeof(char *));
+	io_tlb_orig_addr = alloc_bootmem(io_tlb_nslabs * sizeof(struct swiotlb_phys_addr));
 
 	/*
 	 * Get the overflow emergency buffer

@@ -258,12 +264,12 @@
  		io_tlb_list[i] = IO_TLB_SEGSIZE - OFFSET(i, IO_TLB_SEGSIZE);
 	io_tlb_index = 0;
 
-	io_tlb_orig_addr = (unsigned char **)__get_free_pages(GFP_KERNEL,
-	                           get_order(io_tlb_nslabs * sizeof(char *)));
+	io_tlb_orig_addr = (struct swiotlb_phys_addr *)__get_free_pages(GFP_KERNEL,
+	                           get_order(io_tlb_nslabs * sizeof(struct swiotlb_phys_addr)));
 	if (!io_tlb_orig_addr)
 		goto cleanup3;
 
-	memset(io_tlb_orig_addr, 0, io_tlb_nslabs * sizeof(char *));
+	memset(io_tlb_orig_addr, 0, io_tlb_nslabs * sizeof(struct swiotlb_phys_addr));
 
 	/*
 	 * Get the overflow emergency buffer

@@ -312,20 +318,59 @@
 	return addr >= io_tlb_start && addr < io_tlb_end;
 }
 
+static struct swiotlb_phys_addr swiotlb_bus_to_phys_addr(char *dma_addr)
+{
+	int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT;
+	struct swiotlb_phys_addr buffer = io_tlb_orig_addr[index];
+	buffer.offset += (long)dma_addr & ((1 << IO_TLB_SHIFT) - 1);
+	buffer.page += buffer.offset >> PAGE_SHIFT;
+	buffer.offset &= PAGE_SIZE - 1;
+	return buffer;
+}
+
 static void
-__sync_single(char *buffer, char *dma_addr, size_t size, int dir)
+__sync_single(struct swiotlb_phys_addr buffer, char *dma_addr, size_t size, int dir)
 {
-	if (dir == DMA_TO_DEVICE)
-		memcpy(dma_addr, buffer, size);
-	else
-		memcpy(buffer, dma_addr, size);
+	if (PageHighMem(buffer.page)) {
+		size_t len, bytes;
+		char *dev, *host, *kmp;
+
+		len = size;
+		while (len != 0) {
+			unsigned long flags;
+
+			bytes = len;
+			if ((bytes + buffer.offset) > PAGE_SIZE)
+				bytes = PAGE_SIZE - buffer.offset;
+			local_irq_save(flags); /* protects KM_BOUNCE_READ */
+			kmp  = kmap_atomic(buffer.page, KM_BOUNCE_READ);
+			dev  = dma_addr + size - len;
+			host = kmp + buffer.offset;
+			if (dir == DMA_FROM_DEVICE)
+				memcpy(host, dev, bytes);
+			else
+				memcpy(dev, host, bytes);
+			kunmap_atomic(kmp, KM_BOUNCE_READ);
+			local_irq_restore(flags);
+			len -= bytes;
+			buffer.page++;
+			buffer.offset = 0;
+		}
+	} else {
+		void *v = page_address(buffer.page) + buffer.offset;
+
+		if (dir == DMA_TO_DEVICE)
+			memcpy(dma_addr, v, size);
+		else
+			memcpy(v, dma_addr, size);
+	}
 }
 
 /*
  * Allocates bounce buffer and returns its kernel virtual address.
  */
 static void *
-map_single(struct device *hwdev, char *buffer, size_t size, int dir)
+map_single(struct device *hwdev, struct swiotlb_phys_addr buffer, size_t size, int dir)
 {
 	unsigned long flags;
 	char *dma_addr;

@@ -335,6 +380,7 @@
 	unsigned long mask;
 	unsigned long offset_slots;
 	unsigned long max_slots;
+	struct swiotlb_phys_addr slot_buf;
 
 	mask = dma_get_seg_boundary(hwdev);
 	start_dma_addr = swiotlb_virt_to_bus(io_tlb_start) & mask;

@@ -419,8 +465,13 @@
 	 * This is needed when we sync the memory.  Then we sync the buffer if
 	 * needed.
 	 */
-	for (i = 0; i < nslots; i++)
-		io_tlb_orig_addr[index+i] = buffer + (i << IO_TLB_SHIFT);
+	slot_buf = buffer;
+	for (i = 0; i < nslots; i++) {
+		slot_buf.page += slot_buf.offset >> PAGE_SHIFT;
+		slot_buf.offset &= PAGE_SIZE - 1;
+		io_tlb_orig_addr[index+i] = slot_buf;
+		slot_buf.offset += 1 << IO_TLB_SHIFT;
+	}
 	if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)
 		__sync_single(buffer, dma_addr, size, DMA_TO_DEVICE);
 

@@ -436,12 +487,12 @@
 	unsigned long flags;
 	int i, count, nslots = ALIGN(size, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT;
 	int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT;
-	char *buffer = io_tlb_orig_addr[index];
+	struct swiotlb_phys_addr buffer = swiotlb_bus_to_phys_addr(dma_addr);
 
 	/*
 	 * First, sync the memory before unmapping the entry
 	 */
-	if (buffer && ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL)))
+	if ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL))
 		/*
 		 * bounce... copy the data back into the original buffer * and
 		 * delete the bounce buffer.

@@ -478,10 +529,7 @@
 sync_single(struct device *hwdev, char *dma_addr, size_t size,
 	    int dir, int target)
 {
-	int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT;
-	char *buffer = io_tlb_orig_addr[index];
-
-	buffer += ((unsigned long)dma_addr & ((1 << IO_TLB_SHIFT) - 1));
+	struct swiotlb_phys_addr buffer = swiotlb_bus_to_phys_addr(dma_addr);
 
 	switch (target) {
 	case SYNC_FOR_CPU:

@@ -525,7 +573,10 @@
 		 * swiotlb_map_single(), which will grab memory from
 		 * the lowest available address range.
 		 */
-		ret = map_single(hwdev, NULL, size, DMA_FROM_DEVICE);
+		struct swiotlb_phys_addr buffer;
+		buffer.page = virt_to_page(NULL);
+		buffer.offset = 0;
+		ret = map_single(hwdev, buffer, size, DMA_FROM_DEVICE);
 		if (!ret)
 			return NULL;
 	}

@@ -593,6 +644,7 @@
 {
 	dma_addr_t dev_addr = swiotlb_virt_to_bus(ptr);
 	void *map;
+	struct swiotlb_phys_addr buffer;
 
 	BUG_ON(dir == DMA_NONE);
 	/*

@@ -607,7 +659,9 @@
 	/*
 	 * Oh well, have to allocate and map a bounce buffer.
 	 */
-	map = map_single(hwdev, ptr, size, dir);
+	buffer.page   = virt_to_page(ptr);
+	buffer.offset = (unsigned long)ptr & ~PAGE_MASK;
+	map = map_single(hwdev, buffer, size, dir);
 	if (!map) {
 		swiotlb_full(hwdev, size, dir, 1);
 		map = io_tlb_overflow_buffer;

@@ -752,18 +806,20 @@
 		     int dir, struct dma_attrs *attrs)
 {
 	struct scatterlist *sg;
-	void *addr;
+	struct swiotlb_phys_addr buffer;
 	dma_addr_t dev_addr;
 	int i;
 
 	BUG_ON(dir == DMA_NONE);
 
 	for_each_sg(sgl, sg, nelems, i) {
-		addr = SG_ENT_VIRT_ADDRESS(sg);
-		dev_addr = swiotlb_virt_to_bus(addr);
+		dev_addr = swiotlb_sg_to_bus(sg);
 		if (range_needs_mapping(sg_virt(sg), sg->length) ||
 		    address_needs_mapping(hwdev, dev_addr, sg->length)) {
-			void *map = map_single(hwdev, addr, sg->length, dir);
+			void *map;
+			buffer.page   = sg_page(sg);
+			buffer.offset = sg->offset;
+			map = map_single(hwdev, buffer, sg->length, dir);
 			if (!map) {
 				/* Don't panic here, we expect map_sg users
 				   to do proper error handling. */

@@ -803,11 +859,11 @@
 	BUG_ON(dir == DMA_NONE);
 
 	for_each_sg(sgl, sg, nelems, i) {
-		if (sg->dma_address != SG_ENT_PHYS_ADDRESS(sg))
+		if (sg->dma_address != swiotlb_sg_to_bus(sg))
 			unmap_single(hwdev, swiotlb_bus_to_virt(sg->dma_address),
 				     sg->dma_length, dir);
 		else if (dir == DMA_FROM_DEVICE)
-			dma_mark_clean(SG_ENT_VIRT_ADDRESS(sg), sg->dma_length);
+			dma_mark_clean(swiotlb_bus_to_virt(sg->dma_address), sg->dma_length);
 	}
 }
 EXPORT_SYMBOL(swiotlb_unmap_sg_attrs);

@@ -836,11 +892,11 @@
 	BUG_ON(dir == DMA_NONE);
 
 	for_each_sg(sgl, sg, nelems, i) {
-		if (sg->dma_address != SG_ENT_PHYS_ADDRESS(sg))
+		if (sg->dma_address != swiotlb_sg_to_bus(sg))
 			sync_single(hwdev, swiotlb_bus_to_virt(sg->dma_address),
 				    sg->dma_length, dir, target);
 		else if (dir == DMA_FROM_DEVICE)
-			dma_mark_clean(SG_ENT_VIRT_ADDRESS(sg), sg->dma_length);
+			dma_mark_clean(swiotlb_bus_to_virt(sg->dma_address), sg->dma_length);
 	}
 }
 

@@ -156,6 +156,32 @@
 	return swiotlb_phys_to_bus(page_to_phys(sg_page(sg)) + sg->offset);
 }
 
+static void swiotlb_print_info(unsigned long bytes)
+{
+	phys_addr_t pstart, pend;
+	dma_addr_t bstart, bend;
+
+	pstart = virt_to_phys(io_tlb_start);
+	pend = virt_to_phys(io_tlb_end);
+
+	bstart = swiotlb_phys_to_bus(pstart);
+	bend = swiotlb_phys_to_bus(pend);
+
+	printk(KERN_INFO "Placing %luMB software IO TLB between %p - %p\n",
+	       bytes >> 20, io_tlb_start, io_tlb_end);
+	if (pstart != bstart || pend != bend)
+		printk(KERN_INFO "software IO TLB at phys %#llx - %#llx"
+		       " bus %#llx - %#llx\n",
+		       (unsigned long long)pstart,
+		       (unsigned long long)pend,
+		       (unsigned long long)bstart,
+		       (unsigned long long)bend);
+	else
+		printk(KERN_INFO "software IO TLB at phys %#llx - %#llx\n",
+		       (unsigned long long)pstart,
+		       (unsigned long long)pend);
+}
+
 /*
  * Statically reserve bounce buffer space and initialize bounce buffer data
  * structures for the software IO TLB used to implement the DMA API.

@@ -198,8 +224,7 @@
 	if (!io_tlb_overflow_buffer)
 		panic("Cannot allocate SWIOTLB overflow buffer!\n");
 
-	printk(KERN_INFO "Placing software IO TLB between 0x%lx - 0x%lx\n",
-	       swiotlb_virt_to_bus(io_tlb_start), swiotlb_virt_to_bus(io_tlb_end));
+	swiotlb_print_info(bytes);
 }
 
 void __init

@@ -279,9 +304,7 @@
 	if (!io_tlb_overflow_buffer)
 		goto cleanup4;
 
-	printk(KERN_INFO "Placing %luMB software IO TLB between 0x%lx - "
-	       "0x%lx\n", bytes >> 20,
-	       swiotlb_virt_to_bus(io_tlb_start), swiotlb_virt_to_bus(io_tlb_end));
+	swiotlb_print_info(bytes);
 
 	return 0;
 

@@ -3,6 +3,8 @@
 #include <linux/pci.h>
 #include <linux/cache.h>
 #include <linux/module.h>
+#include <linux/swiotlb.h>
+#include <linux/bootmem.h>
 #include <linux/dma-mapping.h>
 
 #include <asm/iommu.h>

@@ -11,6 +13,16 @@
 
 int swiotlb __read_mostly;
 
+void *swiotlb_alloc_boot(size_t size, unsigned long nslabs)
+{
+	return alloc_bootmem_low_pages(size);
+}
+
+void *swiotlb_alloc(unsigned order, unsigned long nslabs)
+{
+	return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order);
+}
+
 static dma_addr_t
 swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size,
 			int direction)

@@ -66,7 +66,7 @@
 		return dma_ops;
 	else
 		return dev->archdata.dma_ops;
-#endif /* _ASM_X86_DMA_MAPPING_H */
+#endif
 }
 
 /* Make sure we keep the same behaviour */

@@ -84,6 +84,8 @@
 static inline void early_quirks(void) { }
 #endif
 
+extern void pci_iommu_alloc(void);
+
 #endif  /* __KERNEL__ */
 
 #ifdef CONFIG_X86_32

@@ -23,7 +23,6 @@
 			       int reg, int len, u32 value);
 
 extern void dma32_reserve_bootmem(void);
-extern void pci_iommu_alloc(void);
 
 /* The PCI address space does equal the physical memory
  * address space.  The networking and block device layers use

@@ -101,11 +101,15 @@
 	dma32_bootmem_ptr = NULL;
 	dma32_bootmem_size = 0;
 }
+#endif
 
 void __init pci_iommu_alloc(void)
 {
+#ifdef CONFIG_X86_64
 	/* free the range so iommu could get some range less than 4G */
 	dma32_free_bootmem();
+#endif
+
 	/*
 	 * The order of these functions is important for
 	 * fall-back/fail-over reasons

@@ -121,8 +125,6 @@
 	pci_swiotlb_init();
 }
 
-#endif
-
 void *dma_generic_alloc_coherent(struct device *dev, size_t size,
 				 dma_addr_t *dma_addr, gfp_t flag)
 {

@@ -62,8 +62,10 @@
 void __init pci_swiotlb_init(void)
 {
 	/* don't initialize swiotlb if iommu=off (no_iommu=1) */
+#ifdef CONFIG_X86_64
 	if (!iommu_detected && !no_iommu && max_pfn > MAX_DMA32_PFN)
 	       swiotlb = 1;
+#endif
 	if (swiotlb_force)
 		swiotlb = 1;
 	if (swiotlb) {

@@ -21,6 +21,7 @@
 #include <linux/init.h>
 #include <linux/highmem.h>
 #include <linux/pagemap.h>
+#include <linux/pci.h>
 #include <linux/pfn.h>
 #include <linux/poison.h>
 #include <linux/bootmem.h>

@@ -971,6 +972,8 @@
 	int codesize, reservedpages, datasize, initsize;
 	int tmp;
 
+	pci_iommu_alloc();
+
 #ifdef CONFIG_FLATMEM
 	BUG_ON(!mem_map);
 #endif

@@ -23,6 +23,16 @@
 	return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order);
 }
 
+dma_addr_t swiotlb_phys_to_bus(phys_addr_t paddr)
+{
+	return paddr;
+}
+
+phys_addr_t swiotlb_bus_to_phys(dma_addr_t baddr)
+{
+	return baddr;
+}
+
 static dma_addr_t
 swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size,
 			int direction)

@@ -33,6 +33,11 @@
 	return baddr;
 }
 
+int __weak swiotlb_arch_range_needs_mapping(void *ptr, 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)

index db724ba..76821f0 100644
@@ -465,13 +465,9 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 	dma_addr_t dev_addr;
 	void *ret;
 	int order = get_order(size);
-	u64 dma_mask = DMA_32BIT_MASK;
-
-	if (hwdev && hwdev->coherent_dma_mask)
-		dma_mask = hwdev->coherent_dma_mask;
 
 	ret = (void *)__get_free_pages(flags, order);
-	if (ret && !is_buffer_dma_capable(dma_mask, virt_to_bus(ret), size)) {
+	if (ret && !address_needs_mapping(hwdev, virt_to_bus(ret), size)) {
 		/*
 		 * The allocated memory isn't reachable by the device.
 		 * Fall back on swiotlb_map_single().

@@ -495,9 +491,9 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 	dev_addr = virt_to_bus(ret);
 
 	/* Confirm address can be DMA'd by device */
-	if (!is_buffer_dma_capable(dma_mask, dev_addr, size)) {
+	if (!address_needs_mapping(hwdev, dev_addr, size)) {
 		printk("hwdev DMA mask = 0x%016Lx, dev_addr = 0x%016Lx\n",
-		       (unsigned long long)dma_mask,
+		       (unsigned long long)dma_get_mask(hwdev),
 		       (unsigned long long)dev_addr);
 
 		/* DMA_TO_DEVICE to avoid memcpy in unmap_single */

index ed4f44a..e9d5bf6 100644
@@ -118,7 +118,7 @@ static unsigned int io_tlb_index;
  * We need to save away the original address corresponding to a mapped entry
  * for the sync operations.
  */
-static unsigned char **io_tlb_orig_addr;
+static phys_addr_t *io_tlb_orig_addr;
 
 /*
  * Protect the above data structures in the map and unmap calls

@@ -175,7 +175,7 @@ swiotlb_init_with_default_size(size_t default_size)
 	for (i = 0; i < io_tlb_nslabs; i++)
  		io_tlb_list[i] = IO_TLB_SEGSIZE - OFFSET(i, IO_TLB_SEGSIZE);
 	io_tlb_index = 0;
-	io_tlb_orig_addr = alloc_bootmem(io_tlb_nslabs * sizeof(char *));
+	io_tlb_orig_addr = alloc_bootmem(io_tlb_nslabs * sizeof(phys_addr_t));
 
 	/*
 	 * Get the overflow emergency buffer

@@ -253,12 +253,14 @@ swiotlb_late_init_with_default_size(size_t default_size)
  		io_tlb_list[i] = IO_TLB_SEGSIZE - OFFSET(i, IO_TLB_SEGSIZE);
 	io_tlb_index = 0;
 
-	io_tlb_orig_addr = (unsigned char **)__get_free_pages(GFP_KERNEL,
-	                           get_order(io_tlb_nslabs * sizeof(char *)));
+	io_tlb_orig_addr = (phys_addr_t *)
+		__get_free_pages(GFP_KERNEL,
+				 get_order(io_tlb_nslabs *
+					   sizeof(phys_addr_t)));
 	if (!io_tlb_orig_addr)
 		goto cleanup3;
 
-	memset(io_tlb_orig_addr, 0, io_tlb_nslabs * sizeof(char *));
+	memset(io_tlb_orig_addr, 0, io_tlb_nslabs * sizeof(phys_addr_t));
 
 	/*
 	 * Get the overflow emergency buffer

@@ -276,8 +278,8 @@ swiotlb_late_init_with_default_size(size_t default_size)
 	return 0;
 
 cleanup4:
-	free_pages((unsigned long)io_tlb_orig_addr, get_order(io_tlb_nslabs *
-	                                                      sizeof(char *)));
+	free_pages((unsigned long)io_tlb_orig_addr,
+		   get_order(io_tlb_nslabs * sizeof(phys_addr_t)));
 	io_tlb_orig_addr = NULL;
 cleanup3:
 	free_pages((unsigned long)io_tlb_list, get_order(io_tlb_nslabs *

@@ -307,7 +309,7 @@ static int is_swiotlb_buffer(char *addr)
  * Allocates bounce buffer and returns its kernel virtual address.
  */
 static void *
-map_single(struct device *hwdev, char *buffer, size_t size, int dir)
+map_single(struct device *hwdev, phys_addr_t phys, size_t size, int dir)
 {
 	unsigned long flags;
 	char *dma_addr;

@@ -398,9 +400,9 @@ found:
 	 * needed.
 	 */
 	for (i = 0; i < nslots; i++)
-		io_tlb_orig_addr[index+i] = buffer + (i << IO_TLB_SHIFT);
+		io_tlb_orig_addr[index+i] = phys + (i << IO_TLB_SHIFT);
 	if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)
-		memcpy(dma_addr, buffer, size);
+		memcpy(dma_addr, phys_to_virt(phys), size);
 
 	return dma_addr;
 }

@@ -414,17 +416,17 @@ unmap_single(struct device *hwdev, char *dma_addr, size_t size, int dir)
 	unsigned long flags;
 	int i, count, nslots = ALIGN(size, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT;
 	int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT;
-	char *buffer = io_tlb_orig_addr[index];
+	phys_addr_t phys = io_tlb_orig_addr[index];
 
 	/*
 	 * First, sync the memory before unmapping the entry
 	 */
-	if (buffer && ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL)))
+	if (phys && ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL)))
 		/*
 		 * bounce... copy the data back into the original buffer * and
 		 * delete the bounce buffer.
 		 */
-		memcpy(buffer, dma_addr, size);
+		memcpy(phys_to_virt(phys), dma_addr, size);
 
 	/*
 	 * Return the buffer to the free list by setting the corresponding

@@ -457,20 +459,20 @@ sync_single(struct device *hwdev, char *dma_addr, size_t size,
 	    int dir, int target)
 {
 	int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT;
-	char *buffer = io_tlb_orig_addr[index];
+	phys_addr_t phys = io_tlb_orig_addr[index];
 
-	buffer += ((unsigned long)dma_addr & ((1 << IO_TLB_SHIFT) - 1));
+	phys += ((unsigned long)dma_addr & ((1 << IO_TLB_SHIFT) - 1));
 
 	switch (target) {
 	case SYNC_FOR_CPU:
 		if (likely(dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL))
-			memcpy(buffer, dma_addr, size);
+			memcpy(phys_to_virt(phys), dma_addr, size);
 		else
 			BUG_ON(dir != DMA_TO_DEVICE);
 		break;
 	case SYNC_FOR_DEVICE:
 		if (likely(dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL))
-			memcpy(dma_addr, buffer, size);
+			memcpy(dma_addr, phys_to_virt(phys), size);
 		else
 			BUG_ON(dir != DMA_FROM_DEVICE);
 		break;

@@ -509,7 +511,7 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 		 * swiotlb_map_single(), which will grab memory from
 		 * the lowest available address range.
 		 */
-		ret = map_single(hwdev, NULL, size, DMA_FROM_DEVICE);
+		ret = map_single(hwdev, 0, size, DMA_FROM_DEVICE);
 		if (!ret)
 			return NULL;
 	}

@@ -591,7 +593,7 @@ swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size,
 	/*
 	 * Oh well, have to allocate and map a bounce buffer.
 	 */
-	map = map_single(hwdev, ptr, size, dir);
+	map = map_single(hwdev, virt_to_phys(ptr), size, dir);
 	if (!map) {
 		swiotlb_full(hwdev, size, dir, 1);
 		map = io_tlb_overflow_buffer;

@@ -736,18 +738,17 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems,
 		     int dir, struct dma_attrs *attrs)
 {
 	struct scatterlist *sg;
-	void *addr;
 	dma_addr_t dev_addr;
 	int i;
 
 	BUG_ON(dir == DMA_NONE);
 
 	for_each_sg(sgl, sg, nelems, i) {
-		addr = sg_virt(sg);
-		dev_addr = virt_to_dma_addr(hwdev, addr);
+		dev_addr = SG_ENT_BUS_ADDRESS(hwdev, sg);
 		if (swiotlb_force ||
 		    swiotlb_addr_needs_mapping(hwdev, dev_addr, sg->length)) {
-			void *map = map_single(hwdev, addr, sg->length, dir);
+			void *map = map_single(hwdev, sg_phys(sg), sg->length,
+					       dir);
 			if (!map) {
 				/* Don't panic here, we expect map_sg users
 				   to do proper error handling. */

index f4d1be3..ecb5eac 100644
@@ -79,6 +79,16 @@ swiotlb_sync_single_range_for_device(struct device *hwdev, dma_addr_t dev_addr,
 				     unsigned long offset, size_t size,
 				     enum dma_data_direction dir);
 
+extern dma_addr_t
+swiotlb_map_page_attrs(struct device *dev, struct page *page,
+		       unsigned long offset, size_t size,
+		       enum dma_data_direction direction,
+		       struct dma_attrs *attrs);
+
+extern void
+swiotlb_unmap_page_attrs(struct device *hwdev, dma_addr_t dev_addr, size_t size,
+			 enum dma_data_direction dir, struct dma_attrs *attrs);
+
 extern int
 swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr);
 

91 lines not on this page · raw message

index ef09b4c..ed4f44a 100644
@@ -33,10 +33,27 @@
 #include <linux/bootmem.h>
 #include <linux/iommu-helper.h>
 
+
+inline dma_addr_t __weak
+virt_to_dma_addr(struct device *hwdev, void *addr)
+{
+	return virt_to_bus(addr);
+}
+inline void *__weak
+dma_addr_to_virt(struct device *hwdev, dma_addr_t addr)
+{
+	return bus_to_virt(addr);
+}
+inline dma_addr_t __weak
+phys_to_dma_addr(struct device *hwdev, phys_addr_t addr)
+{
+	return (dma_addr_t)addr;
+}
+
 #define OFFSET(val,align) ((unsigned long)	\
 	                   ( (val) & ( (align) - 1)))
 
-#define SG_ENT_BUS_ADDRESS(hwdev, sg)	virt_to_bus(sg_virt(sg))
+#define SG_ENT_BUS_ADDRESS(hwdev, sg) phys_to_dma_addr(hwdev, sg_phys(sg))
 
 /*
  * Maximum allowable number of contiguous slabs to map,

@@ -302,7 +319,7 @@ map_single(struct device *hwdev, char *buffer, size_t size, int dir)
 	unsigned long max_slots;
 
 	mask = dma_get_seg_boundary(hwdev);
-	start_dma_addr = virt_to_bus(io_tlb_start) & mask;
+	start_dma_addr = virt_to_dma_addr(hwdev, io_tlb_start) & mask;
 
 	offset_slots = ALIGN(start_dma_addr, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT;
 	max_slots = mask + 1

@@ -475,7 +492,9 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 		dma_mask = hwdev->coherent_dma_mask;
 
 	ret = (void *)__get_free_pages(flags, order);
-	if (ret && !is_buffer_dma_capable(dma_mask, virt_to_bus(ret), size)) {
+	if (ret && !is_buffer_dma_capable(dma_mask,
+					  virt_to_dma_addr(hwdev, ret),
+					  size)) {
 		/*
 		 * The allocated memory isn't reachable by the device.
 		 * Fall back on swiotlb_map_single().

@@ -496,7 +515,7 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 	}
 
 	memset(ret, 0, size);
-	dev_addr = virt_to_bus(ret);
+	dev_addr = virt_to_dma_addr(hwdev, ret);
 
 	/* Confirm address can be DMA'd by device */
 	if (!is_buffer_dma_capable(dma_mask, dev_addr, size)) {

@@ -556,7 +575,7 @@ dma_addr_t
 swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size,
 			 int dir, struct dma_attrs *attrs)
 {
-	dma_addr_t dev_addr = virt_to_bus(ptr);
+	dma_addr_t dev_addr = virt_to_dma_addr(hwdev, ptr);
 	void *map;
 
 	BUG_ON(dir == DMA_NONE);

@@ -578,7 +597,7 @@ swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size,
 		map = io_tlb_overflow_buffer;
 	}
 
-	dev_addr = virt_to_bus(map);
+	dev_addr = virt_to_dma_addr(hwdev, map);
 
 	/*
 	 * Ensure that the address returned is DMA'ble

@@ -608,7 +627,7 @@ void
 swiotlb_unmap_single_attrs(struct device *hwdev, dma_addr_t dev_addr,
 			   size_t size, int dir, struct dma_attrs *attrs)
 {
-	char *dma_addr = bus_to_virt(dev_addr);
+	char *dma_addr = dma_addr_to_virt(hwdev, dev_addr);
 
 	BUG_ON(dir == DMA_NONE);
 	if (is_swiotlb_buffer(dma_addr))

@@ -638,7 +657,7 @@ static void
 swiotlb_sync_single(struct device *hwdev, dma_addr_t dev_addr,
 		    size_t size, int dir, int target)
 {
-	char *dma_addr = bus_to_virt(dev_addr);
+	char *dma_addr = dma_addr_to_virt(hwdev, dev_addr);
 
 	BUG_ON(dir == DMA_NONE);
 	if (is_swiotlb_buffer(dma_addr))

@@ -669,7 +688,7 @@ swiotlb_sync_single_range(struct device *hwdev, dma_addr_t dev_addr,
 			  unsigned long offset, size_t size,
 			  int dir, int target)
 {
-	char *dma_addr = bus_to_virt(dev_addr) + offset;
+	char *dma_addr = dma_addr_to_virt(hwdev, dev_addr) + offset;
 
 	BUG_ON(dir == DMA_NONE);
 	if (is_swiotlb_buffer(dma_addr))

@@ -725,7 +744,7 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems,
 
 	for_each_sg(sgl, sg, nelems, i) {
 		addr = sg_virt(sg);
-		dev_addr = virt_to_bus(addr);
+		dev_addr = virt_to_dma_addr(hwdev, addr);
 		if (swiotlb_force ||
 		    swiotlb_addr_needs_mapping(hwdev, dev_addr, sg->length)) {
 			void *map = map_single(hwdev, addr, sg->length, dir);

@@ -738,7 +757,7 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems,
 				sgl[0].dma_length = 0;
 				return 0;
 			}
-			sg->dma_address = virt_to_bus(map);
+			sg->dma_address = virt_to_dma_addr(hwdev, map);
 		} else
 			sg->dma_address = dev_addr;
 		sg->dma_length = sg->length;

@@ -769,7 +788,8 @@ swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
 
 	for_each_sg(sgl, sg, nelems, i) {
 		if (sg->dma_address != SG_ENT_BUS_ADDRESS(hwdev, sg))
-			unmap_single(hwdev, bus_to_virt(sg->dma_address),
+			unmap_single(hwdev,
+				     dma_addr_to_virt(hwdev, sg->dma_address),
 				     sg->dma_length, dir);
 		else if (dir == DMA_FROM_DEVICE)
 			dma_mark_clean(sg_virt(sg), sg->dma_length);

@@ -802,7 +822,8 @@ swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
 
 	for_each_sg(sgl, sg, nelems, i) {
 		if (sg->dma_address != SG_ENT_BUS_ADDRESS(hwdev, sg))
-			sync_single(hwdev, bus_to_virt(sg->dma_address),
+			sync_single(hwdev,
+				    dma_addr_to_virt(hwdev, sg->dma_address),
 				    sg->dma_length, dir, target);
 		else if (dir == DMA_FROM_DEVICE)
 			dma_mark_clean(sg_virt(sg), sg->dma_length);

@@ -826,7 +847,7 @@ swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
 int
 swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr)
 {
-	return (dma_addr == virt_to_bus(io_tlb_overflow_buffer));
+	return dma_addr == virt_to_dma_addr(hwdev, io_tlb_overflow_buffer);
 }
 
 /*

@@ -838,7 +859,7 @@ swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr)
 int
 swiotlb_dma_supported(struct device *hwdev, u64 mask)
 {
-	return virt_to_bus(io_tlb_end - 1) <= mask;
+	return virt_to_dma_addr(hwdev, io_tlb_end - 1) <= mask;
 }
 
 EXPORT_SYMBOL(swiotlb_map_single);

index 96e8b10..3d9ac10 100644
@@ -36,8 +36,7 @@
 #define OFFSET(val,align) ((unsigned long)	\
 	                   ( (val) & ( (align) - 1)))
 
-#define SG_ENT_VIRT_ADDRESS(sg)	(sg_virt((sg)))
-#define SG_ENT_PHYS_ADDRESS(sg)	virt_to_bus(SG_ENT_VIRT_ADDRESS(sg))
+#define SG_ENT_PHYS_ADDRESS(sg)	virt_to_bus(sg_virt(sg))
 
 /*
  * Maximum allowable number of contiguous slabs to map,

@@ -721,7 +720,7 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems,
 	BUG_ON(dir == DMA_NONE);
 
 	for_each_sg(sgl, sg, nelems, i) {
-		addr = SG_ENT_VIRT_ADDRESS(sg);
+		addr = sg_virt(sg);
 		dev_addr = virt_to_bus(addr);
 		if (swiotlb_force ||
 		    address_needs_mapping(hwdev, dev_addr, sg->length)) {

@@ -769,7 +768,7 @@ swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
 			unmap_single(hwdev, bus_to_virt(sg->dma_address),
 				     sg->dma_length, dir);
 		else if (dir == DMA_FROM_DEVICE)
-			dma_mark_clean(SG_ENT_VIRT_ADDRESS(sg), sg->dma_length);
+			dma_mark_clean(sg_virt(sg), sg->dma_length);
 	}
 }
 EXPORT_SYMBOL(swiotlb_unmap_sg_attrs);

@@ -802,7 +801,7 @@ swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
 			sync_single(hwdev, bus_to_virt(sg->dma_address),
 				    sg->dma_length, dir, target);
 		else if (dir == DMA_FROM_DEVICE)
-			dma_mark_clean(SG_ENT_VIRT_ADDRESS(sg), sg->dma_length);
+			dma_mark_clean(sg_virt(sg), sg->dma_length);
 	}
 }
 

index 3d9ac10..9b89a9d 100644
@@ -272,8 +272,8 @@ cleanup1:
 	return -ENOMEM;
 }
 
-static int
-address_needs_mapping(struct device *hwdev, dma_addr_t addr, size_t size)
+int __weak
+swiotlb_addr_needs_mapping(struct device *hwdev, dma_addr_t addr, size_t size)
 {
 	return !is_buffer_dma_capable(dma_get_mask(hwdev), addr, size);
 }

@@ -562,7 +562,8 @@ swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size,
 	 * we can safely return the device addr and not worry about bounce
 	 * buffering it.
 	 */
-	if (!address_needs_mapping(hwdev, dev_addr, size) && !swiotlb_force)
+	if (!swiotlb_addr_needs_mapping(hwdev, dev_addr, size) &&
+	    !swiotlb_force)
 		return dev_addr;
 
 	/*

@@ -579,7 +580,7 @@ swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size,
 	/*
 	 * Ensure that the address returned is DMA'ble
 	 */
-	if (address_needs_mapping(hwdev, dev_addr, size))
+	if (swiotlb_addr_needs_mapping(hwdev, dev_addr, size))
 		panic("map_single: bounce buffer is not DMA'ble");
 
 	return dev_addr;

@@ -723,7 +724,7 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems,
 		addr = sg_virt(sg);
 		dev_addr = virt_to_bus(addr);
 		if (swiotlb_force ||
-		    address_needs_mapping(hwdev, dev_addr, sg->length)) {
+		    swiotlb_addr_needs_mapping(hwdev, dev_addr, sg->length)) {
 			void *map = map_single(hwdev, addr, sg->length, dir);
 			if (!map) {
 				/* Don't panic here, we expect map_sg users

index e9d5bf6..ab5d3d7 100644
@@ -14,6 +14,7 @@
  * 04/07/.. ak		Better overflow handling. Assorted fixes.
  * 05/09/10 linville	Add support for syncing ranges, support syncing for
  *			DMA_BIDIRECTIONAL mappings, miscellaneous cleanup.
+ * 08/12/11 beckyb	Add highmem support
  */
 
 #include <linux/cache.h>

@@ -24,6 +25,7 @@
 #include <linux/string.h>
 #include <linux/types.h>
 #include <linux/ctype.h>
+#include <linux/highmem.h>
 
 #include <asm/io.h>
 #include <asm/dma.h>

@@ -306,6 +308,45 @@ static int is_swiotlb_buffer(char *addr)
 }
 
 /*
+ * Bounce: copy the swiotlb buffer back to the original dma location
+ */
+void swiotlb_bounce(phys_addr_t phys, char *dma_addr, size_t size,
+		  enum dma_data_direction dir)
+{
+#ifdef CONFIG_HIGHMEM
+	/* The buffer may not have a mapping.  Map it in and copy */
+	unsigned int offset = ((unsigned long)phys &
+			       ((1 << PAGE_SHIFT) - 1));
+	char *buffer;
+	unsigned int sz = 0;
+	unsigned long flags;
+
+	while (size) {
+		sz = ((PAGE_SIZE - offset) > size) ? size :
+			PAGE_SIZE - offset;
+		local_irq_save(flags);
+		buffer = kmap_atomic(pfn_to_page(phys >> PAGE_SHIFT),
+				     KM_BOUNCE_READ);
+		if (dir == DMA_TO_DEVICE)
+			memcpy(dma_addr, buffer + offset, sz);
+		else
+			memcpy(buffer + offset, dma_addr, sz);
+		kunmap_atomic(buffer, KM_BOUNCE_READ);
+		local_irq_restore(flags);
+		size -= sz;
+		phys += sz;
+		dma_addr += sz;
+		offset = 0;
+	}
+#else
+	if (dir == DMA_TO_DEVICE)
+		memcpy(dma_addr, phys_to_virt(phys), size);
+	else
+		memcpy(phys_to_virt(phys), dma_addr, size);
+#endif
+}
+
+/*
  * Allocates bounce buffer and returns its kernel virtual address.
  */
 static void *

@@ -402,7 +443,7 @@ found:
 	for (i = 0; i < nslots; i++)
 		io_tlb_orig_addr[index+i] = phys + (i << IO_TLB_SHIFT);
 	if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)
-		memcpy(dma_addr, phys_to_virt(phys), size);
+		swiotlb_bounce(phys, dma_addr, size, DMA_TO_DEVICE);
 
 	return dma_addr;
 }

@@ -422,11 +463,7 @@ unmap_single(struct device *hwdev, char *dma_addr, size_t size, int dir)
 	 * First, sync the memory before unmapping the entry
 	 */
 	if (phys && ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL)))
-		/*
-		 * bounce... copy the data back into the original buffer * and
-		 * delete the bounce buffer.
-		 */
-		memcpy(phys_to_virt(phys), dma_addr, size);
+		swiotlb_bounce(phys, dma_addr, size, DMA_FROM_DEVICE);
 
 	/*
 	 * Return the buffer to the free list by setting the corresponding

@@ -466,13 +503,13 @@ sync_single(struct device *hwdev, char *dma_addr, size_t size,
 	switch (target) {
 	case SYNC_FOR_CPU:
 		if (likely(dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL))
-			memcpy(phys_to_virt(phys), dma_addr, size);
+			swiotlb_bounce(phys, dma_addr, size, DMA_FROM_DEVICE);
 		else
 			BUG_ON(dir != DMA_TO_DEVICE);
 		break;
 	case SYNC_FOR_DEVICE:
 		if (likely(dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL))
-			memcpy(dma_addr, phys_to_virt(phys), size);
+			swiotlb_bounce(phys, dma_addr, size, DMA_TO_DEVICE);
 		else
 			BUG_ON(dir != DMA_FROM_DEVICE);
 		break;

index 9b89a9d..add1f92 100644
@@ -36,7 +36,7 @@
 #define OFFSET(val,align) ((unsigned long)	\
 	                   ( (val) & ( (align) - 1)))
 
-#define SG_ENT_PHYS_ADDRESS(sg)	virt_to_bus(sg_virt(sg))
+#define SG_ENT_BUS_ADDRESS(hwdev, sg)	virt_to_bus(sg_virt(sg))
 
 /*
  * Maximum allowable number of contiguous slabs to map,

@@ -765,7 +765,7 @@ swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
 	BUG_ON(dir == DMA_NONE);
 
 	for_each_sg(sgl, sg, nelems, i) {
-		if (sg->dma_address != SG_ENT_PHYS_ADDRESS(sg))
+		if (sg->dma_address != SG_ENT_BUS_ADDRESS(hwdev, sg))
 			unmap_single(hwdev, bus_to_virt(sg->dma_address),
 				     sg->dma_length, dir);
 		else if (dir == DMA_FROM_DEVICE)

@@ -798,7 +798,7 @@ swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
 	BUG_ON(dir == DMA_NONE);
 
 	for_each_sg(sgl, sg, nelems, i) {
-		if (sg->dma_address != SG_ENT_PHYS_ADDRESS(sg))
+		if (sg->dma_address != SG_ENT_BUS_ADDRESS(hwdev, sg))
 			sync_single(hwdev, bus_to_virt(sg->dma_address),
 				    sg->dma_length, dir, target);
 		else if (dir == DMA_FROM_DEVICE)

index add1f92..ef09b4c 100644
@@ -167,8 +167,10 @@ swiotlb_init_with_default_size(size_t default_size)
 	if (!io_tlb_overflow_buffer)
 		panic("Cannot allocate SWIOTLB overflow buffer!\n");
 
-	printk(KERN_INFO "Placing software IO TLB between 0x%lx - 0x%lx\n",
-	       virt_to_bus(io_tlb_start), virt_to_bus(io_tlb_end));
+	printk(KERN_INFO "Placing software IO TLB between cpu physical "
+	       "0x%llx - 0x%llx\n",
+	       (unsigned long long)virt_to_phys(io_tlb_start),
+	       (unsigned long long)virt_to_phys(io_tlb_end));
 }
 
 void __init

@@ -249,9 +251,10 @@ swiotlb_late_init_with_default_size(size_t default_size)
 	if (!io_tlb_overflow_buffer)
 		goto cleanup4;
 
-	printk(KERN_INFO "Placing %luMB software IO TLB between 0x%lx - "
-	       "0x%lx\n", bytes >> 20,
-	       virt_to_bus(io_tlb_start), virt_to_bus(io_tlb_end));
+	printk(KERN_INFO "Placing %luMB software IO TLB between cpu physical "
+	       "0x%llx - 0x%llx\n", bytes >> 20,
+	       (unsigned long long)virt_to_phys(io_tlb_start),
+	       (unsigned long long)virt_to_phys(io_tlb_end));
 
 	return 0;
 

index 1c8a079..f22daf6 100644
@@ -21,7 +21,7 @@ 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)
+		     enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	return intel_map_single(dev, (phys_addr_t)addr, size, dir);
 }

@@ -29,7 +29,7 @@ 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)
+		       enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	intel_unmap_single(dev, iova, size, dir);
 }

@@ -37,7 +37,7 @@ 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)
+		 enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	return intel_map_sg(dev, sglist, nents, dir);
 }

@@ -45,7 +45,8 @@ 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)
+		   int nents, enum dma_data_direction dir,
+		   struct dma_attrs *attrs)
 {
 	intel_unmap_sg(dev, sglist, nents, dir);
 }

index 2769dbf..fd5af38 100644
@@ -91,8 +91,8 @@ hwsw_free_coherent (struct device *dev, size_t size, void *vaddr, dma_addr_t dma
 }
 
 dma_addr_t
-hwsw_map_single_attrs(struct device *dev, void *addr, size_t size, int dir,
-		       struct dma_attrs *attrs)
+hwsw_map_single_attrs(struct device *dev, void *addr, size_t size,
+		      enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
 		return swiotlb_map_single_attrs(dev, addr, size, dir, attrs);

@@ -103,7 +103,7 @@ 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)
+			 enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
 		return swiotlb_unmap_single_attrs(dev, iova, size, dir, attrs);

@@ -114,7 +114,7 @@ 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)
+		   enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
 		return swiotlb_map_sg_attrs(dev, sglist, nents, dir, attrs);

@@ -125,7 +125,7 @@ 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)
+		     enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	if (use_swiotlb(dev))
 		return swiotlb_unmap_sg_attrs(dev, sglist, nents, dir, attrs);

@@ -135,7 +135,8 @@ hwsw_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
 EXPORT_SYMBOL(hwsw_unmap_sg_attrs);
 
 void
-hwsw_sync_single_for_cpu (struct device *dev, dma_addr_t addr, size_t size, int dir)
+hwsw_sync_single_for_cpu (struct device *dev, dma_addr_t addr, size_t size,
+			  enum dma_data_direction dir)
 {
 	if (use_swiotlb(dev))
 		swiotlb_sync_single_for_cpu(dev, addr, size, dir);

@@ -144,7 +145,8 @@ hwsw_sync_single_for_cpu (struct device *dev, dma_addr_t addr, size_t size, int
 }
 
 void
-hwsw_sync_sg_for_cpu (struct device *dev, struct scatterlist *sg, int nelems, int dir)
+hwsw_sync_sg_for_cpu (struct device *dev, struct scatterlist *sg, int nelems,
+		      enum dma_data_direction dir)
 {
 	if (use_swiotlb(dev))
 		swiotlb_sync_sg_for_cpu(dev, sg, nelems, dir);

@@ -153,7 +155,8 @@ hwsw_sync_sg_for_cpu (struct device *dev, struct scatterlist *sg, int nelems, in
 }
 
 void
-hwsw_sync_single_for_device (struct device *dev, dma_addr_t addr, size_t size, int dir)
+hwsw_sync_single_for_device (struct device *dev, dma_addr_t addr, size_t size,
+			     enum dma_data_direction dir)
 {
 	if (use_swiotlb(dev))
 		swiotlb_sync_single_for_device(dev, addr, size, dir);

@@ -162,7 +165,8 @@ hwsw_sync_single_for_device (struct device *dev, dma_addr_t addr, size_t size, i
 }
 
 void
-hwsw_sync_sg_for_device (struct device *dev, struct scatterlist *sg, int nelems, int dir)
+hwsw_sync_sg_for_device (struct device *dev, struct scatterlist *sg, int nelems,
+			 enum dma_data_direction dir)
 {
 	if (use_swiotlb(dev))
 		swiotlb_sync_sg_for_device(dev, sg, nelems, dir);

index d98f0f4..84fe18f 100644
@@ -909,8 +909,8 @@ sba_mark_invalid(struct ioc *ioc, dma_addr_t iova, size_t byte_cnt)
  * See Documentation/DMA-mapping.txt
  */
 dma_addr_t
-sba_map_single_attrs(struct device *dev, void *addr, size_t size, int dir,
-		     struct dma_attrs *attrs)
+sba_map_single_attrs(struct device *dev, void *addr, size_t size,
+		     enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 	dma_addr_t iovp;

@@ -1027,7 +1027,8 @@ sba_mark_clean(struct ioc *ioc, dma_addr_t iova, size_t size)
  * See Documentation/DMA-mapping.txt
  */
 void sba_unmap_single_attrs(struct device *dev, dma_addr_t iova, size_t size,
-			    int dir, struct dma_attrs *attrs)
+			    enum dma_data_direction dir,
+			    struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 #if DELAYED_RESOURCE_CNT > 0

@@ -1423,7 +1424,7 @@ sba_coalesce_chunks(struct ioc *ioc, struct device *dev,
  * See Documentation/DMA-mapping.txt
  */
 int sba_map_sg_attrs(struct device *dev, struct scatterlist *sglist, int nents,
-		     int dir, struct dma_attrs *attrs)
+		     enum dma_data_direction dir, struct dma_attrs *attrs)
 {
 	struct ioc *ioc;
 	int coalesced, filled = 0;

@@ -1515,7 +1516,8 @@ EXPORT_SYMBOL(sba_map_sg_attrs);
  * See Documentation/DMA-mapping.txt
  */
 void sba_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
-			int nents, int dir, struct dma_attrs *attrs)
+			int nents, enum dma_data_direction dir,
+			struct dma_attrs *attrs)
 {
 #ifdef ASSERT_PDIR_SANITY
 	struct ioc *ioc;

index bbab7e2..7452354 100644
@@ -17,32 +17,32 @@ struct dma_mapping_ops {
 	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);
+				size_t size, enum dma_data_direction direction);
 	void            (*unmap_single)(struct device *dev, dma_addr_t addr,
-				size_t size, int direction);
+				size_t size, enum dma_data_direction direction);
 	void            (*sync_single_for_cpu)(struct device *hwdev,
 				dma_addr_t dma_handle, size_t size,
-				int direction);
+				enum dma_data_direction direction);
 	void            (*sync_single_for_device)(struct device *hwdev,
 				dma_addr_t dma_handle, size_t size,
-				int direction);
+				enum dma_data_direction direction);
 	void            (*sync_single_range_for_cpu)(struct device *hwdev,
 				dma_addr_t dma_handle, unsigned long offset,
-				size_t size, int direction);
+				size_t size, enum dma_data_direction direction);
 	void            (*sync_single_range_for_device)(struct device *hwdev,
 				dma_addr_t dma_handle, unsigned long offset,
-				size_t size, int direction);
+				size_t size, enum dma_data_direction direction);
 	void            (*sync_sg_for_cpu)(struct device *hwdev,
 				struct scatterlist *sg, int nelems,
-				int direction);
+				enum dma_data_direction direction);
 	void            (*sync_sg_for_device)(struct device *hwdev,
 				struct scatterlist *sg, int nelems,
-				int direction);
+				enum dma_data_direction direction);
 	int             (*map_sg)(struct device *hwdev, struct scatterlist *sg,
-				int nents, int direction);
+				int nents, enum dma_data_direction direction);
 	void            (*unmap_sg)(struct device *hwdev,
 				struct scatterlist *sg, int nents,
-				int direction);
+				enum dma_data_direction direction);
 	int             (*dma_supported_op)(struct device *hwdev, u64 mask);
 	int		is_phys;
 };

@@ -70,25 +70,26 @@ dma_free_noncoherent(struct device *dev, size_t size, void *cpu_addr,
 }
 #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)
+					size_t size,
+					enum dma_data_direction dir)
 {
 	return dma_map_single_attrs(dev, cpu_addr, size, dir, NULL);
 }
 #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)
+			     int nents, enum dma_data_direction dir)
 {
 	return dma_map_sg_attrs(dev, sgl, nents, dir, NULL);
 }
 #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)
+				    size_t size, enum dma_data_direction dir)
 {
 	return dma_unmap_single_attrs(dev, cpu_addr, size, dir, NULL);
 }
 #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)
+				int nents, enum dma_data_direction dir)
 {
 	return dma_unmap_sg_attrs(dev, sgl, nents, dir, NULL);
 }

index fb79423..81c5629 100644
@@ -6,35 +6,41 @@
 /* SWIOTLB interface */
 
 extern dma_addr_t swiotlb_map_single(struct device *hwdev, void *ptr,
-				     size_t size, int dir);
+				     size_t size,
+				     enum dma_data_direction  dir);
 extern void *swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 				    dma_addr_t *dma_handle, gfp_t flags);
 extern void swiotlb_unmap_single(struct device *hwdev, dma_addr_t dev_addr,
-				 size_t size, int dir);
+				 size_t size,
+				 enum dma_data_direction  dir);
 extern void swiotlb_sync_single_for_cpu(struct device *hwdev,
 					dma_addr_t dev_addr,
-					size_t size, int dir);
+					size_t size,
+					enum dma_data_direction dir);
 extern void swiotlb_sync_single_for_device(struct device *hwdev,
 					   dma_addr_t dev_addr,
-					   size_t size, int dir);
+					   size_t size,
+					   enum dma_data_direction dir);
 extern void swiotlb_sync_single_range_for_cpu(struct device *hwdev,
 					      dma_addr_t dev_addr,
 					      unsigned long offset,
-					      size_t size, int dir);
+					      size_t size,
+					      enum dma_data_direction dir);
 extern void swiotlb_sync_single_range_for_device(struct device *hwdev,
 						 dma_addr_t dev_addr,
 						 unsigned long offset,
-						 size_t size, int dir);
+						 size_t size,
+						 enum dma_data_direction dir);
 extern void swiotlb_sync_sg_for_cpu(struct device *hwdev,
 				    struct scatterlist *sg, int nelems,
-				    int dir);
+				    enum dma_data_direction dir);
 extern void swiotlb_sync_sg_for_device(struct device *hwdev,
 				       struct scatterlist *sg, int nelems,
-				       int dir);
+				       enum dma_data_direction dir);
 extern int swiotlb_map_sg(struct device *hwdev, struct scatterlist *sg,
-			  int nents, int direction);
+			  int nents, enum dma_data_direction direction);
 extern void swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sg,
-			     int nents, int direction);
+			     int nents, enum dma_data_direction direction);
 extern int swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr);
 extern void swiotlb_free_coherent(struct device *hwdev, size_t size,
 				  void *vaddr, dma_addr_t dma_handle);

index 7ccb228..2c0f867 100644
@@ -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();
 }

index 8a924a5..c8c2b04 100644
@@ -60,7 +60,8 @@ sn_default_pci_map(struct pci_dev *pdev, unsigned long paddr, size_t size, int t
 }
 
 static void
-sn_default_pci_unmap(struct pci_dev *pdev, dma_addr_t addr, int direction)
+sn_default_pci_unmap(struct pci_dev *pdev, dma_addr_t addr,
+		     enum dma_data_direction direction)
 {
 	return;
 }

index 53ebb64..87e903d 100644
@@ -174,7 +174,8 @@ EXPORT_SYMBOL(sn_dma_free_coherent);
  *       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,
+				   size_t size,
+				   enum dma_data_direction direction,
 				   struct dma_attrs *attrs)
 {
 	dma_addr_t dma_addr;

@@ -216,7 +217,8 @@ EXPORT_SYMBOL(sn_dma_map_single_attrs);
  * 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,
+			       size_t size,
+			       enum dma_data_direction direction,
 			       struct dma_attrs *attrs)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);

@@ -239,7 +241,7 @@ 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,
+			   int nhwentries, enum dma_data_direction direction,
 			   struct dma_attrs *attrs)
 {
 	int i;

@@ -273,7 +275,8 @@ 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)
+			int nhwentries, enum dma_data_direction direction,
+			struct dma_attrs *attrs)
 {
 	unsigned long phys_addr;
 	struct scatterlist *saved_sg = sgl, *sg;

@@ -323,28 +326,30 @@ int sn_dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
 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)
+				size_t size,
+				enum dma_data_direction direction)
 {
 	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)
+				   size_t size,
+				   enum dma_data_direction direction)
 {
 	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)
+			    int nelems, enum dma_data_direction direction)
 {
 	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)
+			       int nelems, enum dma_data_direction direction)
 {
 	BUG_ON(dev->bus != &pci_bus_type);
 }

index e626e50..c87d57e 100644
@@ -205,7 +205,8 @@ pcibr_dmatrans_direct32(struct pcidev_info * info,
  * DMA mappings for Direct 64 and 32 do not have any DMA maps.
  */
 void
-pcibr_dma_unmap(struct pci_dev *hwdev, dma_addr_t dma_handle, int direction)
+pcibr_dma_unmap(struct pci_dev *hwdev, dma_addr_t dma_handle,
+		enum dma_data_direction direction)
 {
 	struct pcidev_info *pcidev_info = SN_PCIDEV_INFO(hwdev);
 	struct pcibus_info *pcibus_info =

index 7916512..64b46f3 100644
@@ -467,7 +467,8 @@ map_return:
  * resources to release.
  */
 static void
-tioca_dma_unmap(struct pci_dev *pdev, dma_addr_t bus_addr, int dir)
+tioca_dma_unmap(struct pci_dev *pdev, dma_addr_t bus_addr,
+		enum dma_data_direction dir)
 {
 	int i, entry;
 	struct tioca_common *tioca_common;

144 lines not on this page · raw message

index 51fb2c7..c51b25e 100644
@@ -6,35 +6,39 @@
 /* SWIOTLB interface */
 
 extern dma_addr_t swiotlb_map_single(struct device *hwdev, void *ptr,
-				     size_t size, int dir);
+				     size_t size, enum dma_data_direction dir);
 extern void *swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 				    dma_addr_t *dma_handle, gfp_t flags);
 extern void swiotlb_unmap_single(struct device *hwdev, dma_addr_t dev_addr,
-				 size_t size, int dir);
+				 size_t size, enum dma_data_direction dir);
 extern void swiotlb_sync_single_for_cpu(struct device *hwdev,
 					dma_addr_t dev_addr,
-					size_t size, int dir);
+					size_t size,
+					enum dma_data_direction dir);
 extern void swiotlb_sync_single_for_device(struct device *hwdev,
 					   dma_addr_t dev_addr,
-					   size_t size, int dir);
+					   size_t size,
+					   enum dma_data_direction dir);
 extern void swiotlb_sync_single_range_for_cpu(struct device *hwdev,
 					      dma_addr_t dev_addr,
 					      unsigned long offset,
-					      size_t size, int dir);
+					      size_t size,
+					      enum dma_data_direction dir);
 extern void swiotlb_sync_single_range_for_device(struct device *hwdev,
 						 dma_addr_t dev_addr,
 						 unsigned long offset,
-						 size_t size, int dir);
+						 size_t size,
+						 enum dma_data_direction dir);
 extern void swiotlb_sync_sg_for_cpu(struct device *hwdev,
 				    struct scatterlist *sg, int nelems,
-				    int dir);
+				    enum dma_data_direction dir);
 extern void swiotlb_sync_sg_for_device(struct device *hwdev,
 				       struct scatterlist *sg, int nelems,
-				       int dir);
+				       enum dma_data_direction dir);
 extern int swiotlb_map_sg(struct device *hwdev, struct scatterlist *sg,
-			  int nents, int direction);
+			  int nents, enum dma_data_direction direction);
 extern void swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sg,
-			     int nents, int direction);
+			     int nents, enum dma_data_direction direction);
 extern int swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr);
 extern void swiotlb_free_coherent(struct device *hwdev, size_t size,
 				  void *vaddr, dma_addr_t dma_handle);

index 7a6677c..644b60a 100644
@@ -39,7 +39,8 @@ struct iommu_table;
 #define TCE_RPN_MASK     0x0000fffffffff000ULL
 
 extern void tce_build(struct iommu_table *tbl, unsigned long index,
-		      unsigned int npages, unsigned long uaddr, int direction);
+		      unsigned int npages, unsigned long uaddr,
+		      enum dma_data_direction direction);
 extern void tce_free(struct iommu_table *tbl, long index, unsigned int npages);
 extern void * __init alloc_tce_table(void);
 extern void __init free_tce_table(void *tbl);

index ce948aa..13ec5d8 100644
@@ -872,7 +872,7 @@ static dma_addr_t dma_ops_domain_map(struct amd_iommu *iommu,
 				     struct dma_ops_domain *dom,
 				     unsigned long address,
 				     phys_addr_t paddr,
-				     int direction)
+				     enum dma_data_direction direction)
 {
 	u64 *pte, __pte;
 

@@ -932,7 +932,7 @@ static dma_addr_t __map_single(struct device *dev,
 			       struct dma_ops_domain *dma_dom,
 			       phys_addr_t paddr,
 			       size_t size,
-			       int dir,
+			       enum dma_data_direction dir,
 			       bool align,
 			       u64 dma_mask)
 {

@@ -979,7 +979,7 @@ static void __unmap_single(struct amd_iommu *iommu,
 			   struct dma_ops_domain *dma_dom,
 			   dma_addr_t dma_addr,
 			   size_t size,
-			   int dir)
+			   enum dma_data_direction dir)
 {
 	dma_addr_t i, start;
 	unsigned int pages;

@@ -1008,7 +1008,7 @@ 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)
+			     size_t size, enum dma_data_direction dir)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;

@@ -1046,7 +1046,7 @@ 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)
+			 size_t size, enum dma_data_direction dir)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;

@@ -1072,7 +1072,7 @@ static void unmap_single(struct device *dev, dma_addr_t dma_addr,
  * device which is not handled by an AMD IOMMU in the system.
  */
 static int map_sg_no_iommu(struct device *dev, struct scatterlist *sglist,
-			   int nelems, int dir)
+			   int nelems, enum dma_data_direction dir)
 {
 	struct scatterlist *s;
 	int i;

@@ -1090,7 +1090,7 @@ 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)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;

@@ -1152,7 +1152,7 @@ unmap:
  * lists).
  */
 static void unmap_sg(struct device *dev, struct scatterlist *sglist,
-		     int nelems, int dir)
+		     int nelems, enum dma_data_direction dir)
 {
 	unsigned long flags;
 	struct amd_iommu *iommu;

index d28bbdc..76a55c0 100644
@@ -297,7 +297,8 @@ static unsigned long iommu_range_alloc(struct device *dev,
 }
 
 static dma_addr_t iommu_alloc(struct device *dev, struct iommu_table *tbl,
-			      void *vaddr, unsigned int npages, int direction)
+			      void *vaddr, unsigned int npages,
+			      enum dma_data_direction direction)
 {
 	unsigned long entry;
 	dma_addr_t ret = bad_dma_address;

@@ -381,7 +382,7 @@ static inline struct iommu_table *find_iommu_table(struct device *dev)
 }
 
 static void calgary_unmap_sg(struct device *dev,
-	struct scatterlist *sglist, int nelems, int direction)
+	struct scatterlist *sglist, int nelems, enum dma_data_direction dir)
 {
 	struct iommu_table *tbl = find_iommu_table(dev);
 	struct scatterlist *s;

@@ -404,7 +405,7 @@ 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 direction)
 {
 	struct iommu_table *tbl = find_iommu_table(dev);
 	struct scatterlist *s;

@@ -446,7 +447,7 @@ error:
 }
 
 static dma_addr_t calgary_map_single(struct device *dev, phys_addr_t paddr,
-	size_t size, int direction)
+	size_t size, enum dma_data_direction direction)
 {
 	void *vaddr = phys_to_virt(paddr);
 	unsigned long uaddr;

@@ -460,7 +461,7 @@ static dma_addr_t calgary_map_single(struct device *dev, phys_addr_t paddr,
 }
 
 static void calgary_unmap_single(struct device *dev, dma_addr_t dma_handle,
-	size_t size, int direction)
+	size_t size, enum dma_data_direction direction)
 {
 	struct iommu_table *tbl = find_iommu_table(dev);
 	unsigned int npages;

index ba7ad83..f3cd99b 100644
@@ -188,7 +188,8 @@ static void dump_leak(void)
 # define CLEAR_LEAK(x)
 #endif
 
-static void iommu_full(struct device *dev, size_t size, int dir)
+static void
+iommu_full(struct device *dev, size_t size, enum dma_data_direction dir)
 {
 	/*
 	 * Ran out of IOMMU space for this operation. This is very bad.

@@ -231,7 +232,8 @@ nonforced_iommu(struct device *dev, unsigned long addr, size_t size)
  * Caller needs to check if the iommu is needed and flush.
  */
 static dma_addr_t dma_map_area(struct device *dev, dma_addr_t phys_mem,
-				size_t size, int dir, unsigned long align_mask)
+			       size_t size, enum dma_data_direction dir,
+			       unsigned long align_mask)
 {
 	unsigned long npages = iommu_num_pages(phys_mem, size, PAGE_SIZE);
 	unsigned long iommu_page = alloc_iommu(dev, npages, align_mask);

@@ -256,7 +258,8 @@ 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)
+gart_map_single(struct device *dev, phys_addr_t paddr, size_t size,
+		enum dma_data_direction dir)
 {
 	unsigned long bus;
 

@@ -276,7 +279,7 @@ 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)
+			      size_t size, enum dma_data_direction direction)
 {
 	unsigned long iommu_page;
 	int npages;

@@ -299,7 +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)
+gart_unmap_sg(struct device *dev, struct scatterlist *sg, int nents,
+	      enum dma_data_direction dir)
 {
 	struct scatterlist *s;
 	int i;

@@ -313,7 +317,7 @@ gart_unmap_sg(struct device *dev, struct scatterlist *sg, int nents, int dir)
 
 /* Fallback for dma_map_sg in case of overflow */
 static int dma_map_sg_nonforce(struct device *dev, struct scatterlist *sg,
-			       int nents, int dir)
+			       int nents, enum dma_data_direction dir)
 {
 	struct scatterlist *s;
 	int i;

@@ -401,7 +405,8 @@ dma_map_cont(struct device *dev, struct scatterlist *start, int nelems,
  * 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)
+gart_map_sg(struct device *dev, struct scatterlist *sg, int nents,
+	    enum dma_data_direction dir)
 {
 	struct scatterlist *s, *ps, *start_sg, *sgmap;
 	int need = 0, nextneed, i, out, start;

index c70ab5a..a4c6b7f 100644
@@ -27,7 +27,7 @@ check_addr(char *name, struct device *hwdev, dma_addr_t bus, size_t size)
 
 static dma_addr_t
 nommu_map_single(struct device *hwdev, phys_addr_t paddr, size_t size,
-	       int direction)
+		 enum dma_data_direction direction)
 {
 	dma_addr_t bus = paddr;
 	WARN_ON(size == 0);

@@ -54,7 +54,7 @@ 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 direction)
 {
 	struct scatterlist *s;
 	int i;

index 3c539d1..c400a0b 100644
@@ -13,7 +13,7 @@ int swiotlb __read_mostly;
 
 static dma_addr_t
 swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size,
-			int direction)
+			enum dma_data_direction direction)
 {
 	return swiotlb_map_single(hwdev, phys_to_virt(paddr), size, direction);
 }

index 9e540fe..29bc5b6 100644
@@ -46,7 +46,8 @@ static inline void flush_tce(void* tceaddr)
 }
 
 void tce_build(struct iommu_table *tbl, unsigned long index,
-	unsigned int npages, unsigned long uaddr, int direction)
+	unsigned int npages, unsigned long uaddr,
+	enum dma_data_direction direction)
 {
 	u64* tp;
 	u64 t;

index 213a5c8..0232293 100644
@@ -1997,7 +1997,8 @@ get_valid_domain_for_dev(struct pci_dev *pdev)
 }
 
 static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr,
-				     size_t size, int dir, u64 dma_mask)
+				     size_t size,
+				     enum dma_data_direction dir, u64 dma_mask)
 {
 	struct pci_dev *pdev = to_pci_dev(hwdev);
 	struct dmar_domain *domain;