Merge branch 'for-v3.9' of git://git.linaro.org/people/mszyprowski/linux-dma-mapping

Pull DMA-mapping updates from Marek Szyprowski:
 "This time all patches are related only to ARM DMA-mapping subsystem.
  The main extension provided by this pull request is highmem support.
  Besides that it contains a bunch of small bugfixes and cleanups."

* 'for-v3.9' of git://git.linaro.org/people/mszyprowski/linux-dma-mapping:
  ARM: DMA-mapping: fix memory leak in IOMMU dma-mapping implementation
  ARM: dma-mapping: Add maximum alignment order for dma iommu buffers
  ARM: dma-mapping: use himem for DMA buffers for IOMMU-mapped devices
  ARM: dma-mapping: add support for CMA regions placed in highmem zone
  arm: dma mapping: export arm iommu functions
  ARM: dma-mapping: Add arm_iommu_detach_device()
  ARM: dma-mapping: Add macro to_dma_iommu_mapping()
  ARM: dma-mapping: Set arm_dma_set_mask() for iommu->set_dma_mask()
  ARM: iommu: Include linux/kref.h in asm/dma-iommu.h

arch/arm/Kconfig

@@ -77,6 +77,27 @@ config ARM_DMA_USE_IOMMU
 	select ARM_HAS_SG_CHAIN
 	select NEED_SG_DMA_LENGTH
 
+if ARM_DMA_USE_IOMMU
+
+config ARM_DMA_IOMMU_ALIGNMENT
+	int "Maximum PAGE_SIZE order of alignment for DMA IOMMU buffers"
+	range 4 9
+	default 8
+	help
+	  DMA mapping framework by default aligns all buffers to the smallest
+	  PAGE_SIZE order which is greater than or equal to the requested buffer
+	  size. This works well for buffers up to a few hundreds kilobytes, but
+	  for larger buffers it just a waste of address space. Drivers which has
+	  relatively small addressing window (like 64Mib) might run out of
+	  virtual space with just a few allocations.
+
+	  With this parameter you can specify the maximum PAGE_SIZE order for
+	  DMA IOMMU buffers. Larger buffers will be aligned only to this
+	  specified order. The order is expressed as a power of two multiplied
+	  by the PAGE_SIZE.
+
+endif
+
 config HAVE_PWM
 	bool
 

arch/arm/include/asm/device.h

@@ -27,4 +27,10 @@ struct pdev_archdata {
 #endif
 };
 
+#ifdef CONFIG_ARM_DMA_USE_IOMMU
+#define to_dma_iommu_mapping(dev) ((dev)->archdata.mapping)
+#else
+#define to_dma_iommu_mapping(dev) NULL
+#endif
+
 #endif

arch/arm/include/asm/dma-iommu.h

@@ -7,6 +7,7 @@
 #include <linux/scatterlist.h>
 #include <linux/dma-debug.h>
 #include <linux/kmemcheck.h>
+#include <linux/kref.h>
 
 struct dma_iommu_mapping {
 	/* iommu specific data */
@@ -29,6 +30,7 @@ void arm_iommu_release_mapping(struct dma_iommu_mapping *mapping);
 
 int arm_iommu_attach_device(struct device *dev,
 					struct dma_iommu_mapping *mapping);
+void arm_iommu_detach_device(struct device *dev);
 
 #endif /* __KERNEL__ */
 #endif

arch/arm/mm/dma-mapping.c

@@ -186,13 +186,24 @@ static u64 get_coherent_dma_mask(struct device *dev)
 
 static void __dma_clear_buffer(struct page *page, size_t size)
 {
-	void *ptr;
 	/*
 	 * Ensure that the allocated pages are zeroed, and that any data
 	 * lurking in the kernel direct-mapped region is invalidated.
 	 */
-	ptr = page_address(page);
+	if (PageHighMem(page)) {
-	if (ptr) {
+		phys_addr_t base = __pfn_to_phys(page_to_pfn(page));
+		phys_addr_t end = base + size;
+		while (size > 0) {
+			void *ptr = kmap_atomic(page);
+			memset(ptr, 0, PAGE_SIZE);
+			dmac_flush_range(ptr, ptr + PAGE_SIZE);
+			kunmap_atomic(ptr);
+			page++;
+			size -= PAGE_SIZE;
+		}
+		outer_flush_range(base, end);
+	} else {
+		void *ptr = page_address(page);
 		memset(ptr, 0, size);
 		dmac_flush_range(ptr, ptr + size);
 		outer_flush_range(__pa(ptr), __pa(ptr) + size);
@@ -243,7 +254,8 @@ static void __dma_free_buffer(struct page *page, size_t size)
 #endif
 
 static void *__alloc_from_contiguous(struct device *dev, size_t size,
-				     pgprot_t prot, struct page **ret_page);
+				     pgprot_t prot, struct page **ret_page,
+				     const void *caller);
 
 static void *__alloc_remap_buffer(struct device *dev, size_t size, gfp_t gfp,
 				 pgprot_t prot, struct page **ret_page,
@@ -346,10 +358,11 @@ static int __init atomic_pool_init(void)
 		goto no_pages;
 
 	if (IS_ENABLED(CONFIG_CMA))
-		ptr = __alloc_from_contiguous(NULL, pool->size, prot, &page);
+		ptr = __alloc_from_contiguous(NULL, pool->size, prot, &page,
+					      atomic_pool_init);
 	else
 		ptr = __alloc_remap_buffer(NULL, pool->size, GFP_KERNEL, prot,
-					   &page, NULL);
+					   &page, atomic_pool_init);
 	if (ptr) {
 		int i;
 
@@ -542,27 +555,41 @@ static int __free_from_pool(void *start, size_t size)
 }
 
 static void *__alloc_from_contiguous(struct device *dev, size_t size,
-				     pgprot_t prot, struct page **ret_page)
+				     pgprot_t prot, struct page **ret_page,
+				     const void *caller)
 {
 	unsigned long order = get_order(size);
 	size_t count = size >> PAGE_SHIFT;
 	struct page *page;
+	void *ptr;
 
 	page = dma_alloc_from_contiguous(dev, count, order);
 	if (!page)
 		return NULL;
 
 	__dma_clear_buffer(page, size);
-	__dma_remap(page, size, prot);
 
+	if (PageHighMem(page)) {
+		ptr = __dma_alloc_remap(page, size, GFP_KERNEL, prot, caller);
+		if (!ptr) {
+			dma_release_from_contiguous(dev, page, count);
+			return NULL;
+		}
+	} else {
+		__dma_remap(page, size, prot);
+		ptr = page_address(page);
+	}
 	*ret_page = page;
-	return page_address(page);
+	return ptr;
 }
 
 static void __free_from_contiguous(struct device *dev, struct page *page,
-				   size_t size)
+				   void *cpu_addr, size_t size)
 {
-	__dma_remap(page, size, pgprot_kernel);
+	if (PageHighMem(page))
+		__dma_free_remap(cpu_addr, size);
+	else
+		__dma_remap(page, size, pgprot_kernel);
 	dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);
 }
 
@@ -583,9 +610,9 @@ static inline pgprot_t __get_dma_pgprot(struct dma_attrs *attrs, pgprot_t prot)
 #define __get_dma_pgprot(attrs, prot)	__pgprot(0)
 #define __alloc_remap_buffer(dev, size, gfp, prot, ret, c)	NULL
 #define __alloc_from_pool(size, ret_page)			NULL
-#define __alloc_from_contiguous(dev, size, prot, ret)		NULL
+#define __alloc_from_contiguous(dev, size, prot, ret, c)	NULL
 #define __free_from_pool(cpu_addr, size)			0
-#define __free_from_contiguous(dev, page, size)			do { } while (0)
+#define __free_from_contiguous(dev, page, cpu_addr, size)	do { } while (0)
 #define __dma_free_remap(cpu_addr, size)			do { } while (0)
 
 #endif	/* CONFIG_MMU */
@@ -645,7 +672,7 @@ static void *__dma_alloc(struct device *dev, size_t size, dma_addr_t *handle,
 	else if (!IS_ENABLED(CONFIG_CMA))
 		addr = __alloc_remap_buffer(dev, size, gfp, prot, &page, caller);
 	else
-		addr = __alloc_from_contiguous(dev, size, prot, &page);
+		addr = __alloc_from_contiguous(dev, size, prot, &page, caller);
 
 	if (addr)
 		*handle = pfn_to_dma(dev, page_to_pfn(page));
@@ -739,7 +766,7 @@ static void __arm_dma_free(struct device *dev, size_t size, void *cpu_addr,
 		 * Non-atomic allocations cannot be freed with IRQs disabled
 		 */
 		WARN_ON(irqs_disabled());
-		__free_from_contiguous(dev, page, size);
+		__free_from_contiguous(dev, page, cpu_addr, size);
 	}
 }
 
@@ -1002,6 +1029,9 @@ static inline dma_addr_t __alloc_iova(struct dma_iommu_mapping *mapping,
 	unsigned int count, start;
 	unsigned long flags;
 
+	if (order > CONFIG_ARM_DMA_IOMMU_ALIGNMENT)
+		order = CONFIG_ARM_DMA_IOMMU_ALIGNMENT;
+
 	count = ((PAGE_ALIGN(size) >> PAGE_SHIFT) +
 		 (1 << mapping->order) - 1) >> mapping->order;
 
@@ -1068,12 +1098,17 @@ static struct page **__iommu_alloc_buffer(struct device *dev, size_t size,
 		return pages;
 	}
 
+	/*
+	 * IOMMU can map any pages, so himem can also be used here
+	 */
+	gfp |= __GFP_NOWARN | __GFP_HIGHMEM;
+
 	while (count) {
 		int j, order = __fls(count);
 
-		pages[i] = alloc_pages(gfp | __GFP_NOWARN, order);
+		pages[i] = alloc_pages(gfp, order);
 		while (!pages[i] && order)
-			pages[i] = alloc_pages(gfp | __GFP_NOWARN, --order);
+			pages[i] = alloc_pages(gfp, --order);
 		if (!pages[i])
 			goto error;
 
@@ -1257,11 +1292,11 @@ err_mapping:
 	return NULL;
 }
 
-static void __iommu_free_atomic(struct device *dev, struct page **pages,
+static void __iommu_free_atomic(struct device *dev, void *cpu_addr,
 				dma_addr_t handle, size_t size)
 {
 	__iommu_remove_mapping(dev, handle, size);
-	__free_from_pool(page_address(pages[0]), size);
+	__free_from_pool(cpu_addr, size);
 }
 
 static void *arm_iommu_alloc_attrs(struct device *dev, size_t size,
@@ -1344,7 +1379,7 @@ void arm_iommu_free_attrs(struct device *dev, size_t size, void *cpu_addr,
 	}
 
 	if (__in_atomic_pool(cpu_addr, size)) {
-		__iommu_free_atomic(dev, pages, handle, size);
+		__iommu_free_atomic(dev, cpu_addr, handle, size);
 		return;
 	}
 
@@ -1732,6 +1767,8 @@ struct dma_map_ops iommu_ops = {
 	.unmap_sg		= arm_iommu_unmap_sg,
 	.sync_sg_for_cpu	= arm_iommu_sync_sg_for_cpu,
 	.sync_sg_for_device	= arm_iommu_sync_sg_for_device,
+
+	.set_dma_mask		= arm_dma_set_mask,
 };
 
 struct dma_map_ops iommu_coherent_ops = {
@@ -1745,6 +1782,8 @@ struct dma_map_ops iommu_coherent_ops = {
 
 	.map_sg		= arm_coherent_iommu_map_sg,
 	.unmap_sg	= arm_coherent_iommu_unmap_sg,
+
+	.set_dma_mask	= arm_dma_set_mask,
 };
 
 /**
@@ -1799,6 +1838,7 @@ err2:
 err:
 	return ERR_PTR(err);
 }
+EXPORT_SYMBOL_GPL(arm_iommu_create_mapping);
 
 static void release_iommu_mapping(struct kref *kref)
 {
@@ -1815,6 +1855,7 @@ void arm_iommu_release_mapping(struct dma_iommu_mapping *mapping)
 	if (mapping)
 		kref_put(&mapping->kref, release_iommu_mapping);
 }
+EXPORT_SYMBOL_GPL(arm_iommu_release_mapping);
 
 /**
  * arm_iommu_attach_device
@@ -1843,5 +1884,32 @@ int arm_iommu_attach_device(struct device *dev,
 	pr_debug("Attached IOMMU controller to %s device.\n", dev_name(dev));
 	return 0;
 }
+EXPORT_SYMBOL_GPL(arm_iommu_attach_device);
+
+/**
+ * arm_iommu_detach_device
+ * @dev: valid struct device pointer
+ *
+ * Detaches the provided device from a previously attached map.
+ * This voids the dma operations (dma_map_ops pointer)
+ */
+void arm_iommu_detach_device(struct device *dev)
+{
+	struct dma_iommu_mapping *mapping;
+
+	mapping = to_dma_iommu_mapping(dev);
+	if (!mapping) {
+		dev_warn(dev, "Not attached\n");
+		return;
+	}
+
+	iommu_detach_device(mapping->domain, dev);
+	kref_put(&mapping->kref, release_iommu_mapping);
+	mapping = NULL;
+	set_dma_ops(dev, NULL);
+
+	pr_debug("Detached IOMMU controller from %s device.\n", dev_name(dev));
+}
+EXPORT_SYMBOL_GPL(arm_iommu_detach_device);
 
 #endif