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: (61 commits)
  amd-iommu: remove unnecessary "AMD IOMMU: " prefix
  amd-iommu: detach device explicitly before attaching it to a new domain
  amd-iommu: remove BUS_NOTIFY_BOUND_DRIVER handling
  dma-debug: simplify logic in driver_filter()
  dma-debug: disable/enable irqs only once in device_dma_allocations
  dma-debug: use pr_* instead of printk(KERN_* ...)
  dma-debug: code style fixes
  dma-debug: comment style fixes
  dma-debug: change hash_bucket_find from first-fit to best-fit
  x86: enable GART-IOMMU only after setting up protection methods
  amd_iommu: fix lock imbalance
  dma-debug: add documentation for the driver filter
  dma-debug: add dma_debug_driver kernel command line
  dma-debug: add debugfs file for driver filter
  dma-debug: add variables and checks for driver filter
  dma-debug: fix debug_dma_sync_sg_for_cpu and debug_dma_sync_sg_for_device
  dma-debug: use sg_dma_len accessor
  dma-debug: use sg_dma_address accessor instead of using dma_address directly
  amd-iommu: don't free dma adresses below 512MB with CONFIG_IOMMU_STRESS
  amd-iommu: don't preallocate page tables with CONFIG_IOMMU_STRESS
  ...

2 files changed, 449 insertions, 102 deletions

diff --git a/lib/dma-debug.c b/lib/dma-debug.c
index 69da09a..ad65fc0 100644
--- a/lib/dma-debug.c
+++ b/lib/dma-debug.c
@@ -23,9 +23,11 @@
 #include <linux/dma-debug.h>
 #include <linux/spinlock.h>
 #include <linux/debugfs.h>
+#include <linux/uaccess.h>
 #include <linux/device.h>
 #include <linux/types.h>
 #include <linux/sched.h>
+#include <linux/ctype.h>
 #include <linux/list.h>
 #include <linux/slab.h>
 
@@ -85,6 +87,7 @@ static u32 show_num_errors = 1;
 
 static u32 num_free_entries;
 static u32 min_free_entries;
+static u32 nr_total_entries;
 
 /* number of preallocated entries requested by kernel cmdline */
 static u32 req_entries;
@@ -97,6 +100,16 @@ static struct dentry *show_all_errors_dent  __read_mostly;
 static struct dentry *show_num_errors_dent  __read_mostly;
 static struct dentry *num_free_entries_dent __read_mostly;
 static struct dentry *min_free_entries_dent __read_mostly;
+static struct dentry *filter_dent           __read_mostly;
+
+/* per-driver filter related state */
+
+#define NAME_MAX_LEN	64
+
+static char                  current_driver_name[NAME_MAX_LEN] __read_mostly;
+static struct device_driver *current_driver                    __read_mostly;
+
+static DEFINE_RWLOCK(driver_name_lock);
 
 static const char *type2name[4] = { "single", "page",
 				    "scather-gather", "coherent" };
@@ -104,6 +117,11 @@ static const char *type2name[4] = { "single", "page",
 static const char *dir2name[4] = { "DMA_BIDIRECTIONAL", "DMA_TO_DEVICE",
 				   "DMA_FROM_DEVICE", "DMA_NONE" };
 
+/* little merge helper - remove it after the merge window */
+#ifndef BUS_NOTIFY_UNBOUND_DRIVER
+#define BUS_NOTIFY_UNBOUND_DRIVER 0x0005
+#endif
+
 /*
  * The access to some variables in this macro is racy. We can't use atomic_t
  * here because all these variables are exported to debugfs. Some of them even
@@ -121,15 +139,54 @@ static inline void dump_entry_trace(struct dma_debug_entry *entry)
 {
 #ifdef CONFIG_STACKTRACE
 	if (entry) {
-		printk(KERN_WARNING "Mapped at:\n");
+		pr_warning("Mapped at:\n");
 		print_stack_trace(&entry->stacktrace, 0);
 	}
 #endif
 }
 
+static bool driver_filter(struct device *dev)
+{
+	struct device_driver *drv;
+	unsigned long flags;
+	bool ret;
+
+	/* driver filter off */
+	if (likely(!current_driver_name[0]))
+		return true;
+
+	/* driver filter on and initialized */
+	if (current_driver && dev->driver == current_driver)
+		return true;
+
+	if (current_driver || !current_driver_name[0])
+		return false;
+
+	/* driver filter on but not yet initialized */
+	drv = get_driver(dev->driver);
+	if (!drv)
+		return false;
+
+	/* lock to protect against change of current_driver_name */
+	read_lock_irqsave(&driver_name_lock, flags);
+
+	ret = false;
+	if (drv->name &&
+	    strncmp(current_driver_name, drv->name, NAME_MAX_LEN - 1) == 0) {
+		current_driver = drv;
+		ret = true;
+	}
+
+	read_unlock_irqrestore(&driver_name_lock, flags);
+	put_driver(drv);
+
+	return ret;
+}
+
 #define err_printk(dev, entry, format, arg...) do {		\
 		error_count += 1;				\
-		if (show_all_errors || show_num_errors > 0) {	\
+		if (driver_filter(dev) &&			\
+		    (show_all_errors || show_num_errors > 0)) {	\
 			WARN(1, "%s %s: " format,		\
 			     dev_driver_string(dev),		\
 			     dev_name(dev) , ## arg);		\
@@ -185,15 +242,50 @@ static void put_hash_bucket(struct hash_bucket *bucket,
 static struct dma_debug_entry *hash_bucket_find(struct hash_bucket *bucket,
 						struct dma_debug_entry *ref)
 {
-	struct dma_debug_entry *entry;
+	struct dma_debug_entry *entry, *ret = NULL;
+	int matches = 0, match_lvl, last_lvl = 0;
 
 	list_for_each_entry(entry, &bucket->list, list) {
-		if ((entry->dev_addr == ref->dev_addr) &&
-		    (entry->dev == ref->dev))
+		if ((entry->dev_addr != ref->dev_addr) ||
+		    (entry->dev != ref->dev))
+			continue;
+
+		/*
+		 * Some drivers map the same physical address multiple
+		 * times. Without a hardware IOMMU this results in the
+		 * same device addresses being put into the dma-debug
+		 * hash multiple times too. This can result in false
+		 * positives being reported. Therfore we implement a
+		 * best-fit algorithm here which returns the entry from
+		 * the hash which fits best to the reference value
+		 * instead of the first-fit.
+		 */
+		matches += 1;
+		match_lvl = 0;
+		entry->size      == ref->size      ? ++match_lvl : match_lvl;
+		entry->type      == ref->type      ? ++match_lvl : match_lvl;
+		entry->direction == ref->direction ? ++match_lvl : match_lvl;
+
+		if (match_lvl == 3) {
+			/* perfect-fit - return the result */
 			return entry;
+		} else if (match_lvl > last_lvl) {
+			/*
+			 * We found an entry that fits better then the
+			 * previous one
+			 */
+			last_lvl = match_lvl;
+			ret      = entry;
+		}
 	}
 
-	return NULL;
+	/*
+	 * If we have multiple matches but no perfect-fit, just return
+	 * NULL.
+	 */
+	ret = (matches == 1) ? ret : NULL;
+
+	return ret;
 }
 
 /*
@@ -257,6 +349,21 @@ static void add_dma_entry(struct dma_debug_entry *entry)
 	put_hash_bucket(bucket, &flags);
 }
 
+static struct dma_debug_entry *__dma_entry_alloc(void)
+{
+	struct dma_debug_entry *entry;
+
+	entry = list_entry(free_entries.next, struct dma_debug_entry, list);
+	list_del(&entry->list);
+	memset(entry, 0, sizeof(*entry));
+
+	num_free_entries -= 1;
+	if (num_free_entries < min_free_entries)
+		min_free_entries = num_free_entries;
+
+	return entry;
+}
+
 /* struct dma_entry allocator
  *
  * The next two functions implement the allocator for
@@ -270,15 +377,12 @@ static struct dma_debug_entry *dma_entry_alloc(void)
 	spin_lock_irqsave(&free_entries_lock, flags);
 
 	if (list_empty(&free_entries)) {
-		printk(KERN_ERR "DMA-API: debugging out of memory "
-				"- disabling\n");
+		pr_err("DMA-API: debugging out of memory - disabling\n");
 		global_disable = true;
 		goto out;
 	}
 
-	entry = list_entry(free_entries.next, struct dma_debug_entry, list);
-	list_del(&entry->list);
-	memset(entry, 0, sizeof(*entry));
+	entry = __dma_entry_alloc();
 
 #ifdef CONFIG_STACKTRACE
 	entry->stacktrace.max_entries = DMA_DEBUG_STACKTRACE_ENTRIES;
@@ -286,9 +390,6 @@ static struct dma_debug_entry *dma_entry_alloc(void)
 	entry->stacktrace.skip = 2;
 	save_stack_trace(&entry->stacktrace);
 #endif
-	num_free_entries -= 1;
-	if (num_free_entries < min_free_entries)
-		min_free_entries = num_free_entries;
 
 out:
 	spin_unlock_irqrestore(&free_entries_lock, flags);
@@ -310,6 +411,53 @@ static void dma_entry_free(struct dma_debug_entry *entry)
 	spin_unlock_irqrestore(&free_entries_lock, flags);
 }
 
+int dma_debug_resize_entries(u32 num_entries)
+{
+	int i, delta, ret = 0;
+	unsigned long flags;
+	struct dma_debug_entry *entry;
+	LIST_HEAD(tmp);
+
+	spin_lock_irqsave(&free_entries_lock, flags);
+
+	if (nr_total_entries < num_entries) {
+		delta = num_entries - nr_total_entries;
+
+		spin_unlock_irqrestore(&free_entries_lock, flags);
+
+		for (i = 0; i < delta; i++) {
+			entry = kzalloc(sizeof(*entry), GFP_KERNEL);
+			if (!entry)
+				break;
+
+			list_add_tail(&entry->list, &tmp);
+		}
+
+		spin_lock_irqsave(&free_entries_lock, flags);
+
+		list_splice(&tmp, &free_entries);
+		nr_total_entries += i;
+		num_free_entries += i;
+	} else {
+		delta = nr_total_entries - num_entries;
+
+		for (i = 0; i < delta && !list_empty(&free_entries); i++) {
+			entry = __dma_entry_alloc();
+			kfree(entry);
+		}
+
+		nr_total_entries -= i;
+	}
+
+	if (nr_total_entries != num_entries)
+		ret = 1;
+
+	spin_unlock_irqrestore(&free_entries_lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(dma_debug_resize_entries);
+
 /*
  * DMA-API debugging init code
  *
@@ -334,8 +482,7 @@ static int prealloc_memory(u32 num_entries)
 	num_free_entries = num_entries;
 	min_free_entries = num_entries;
 
-	printk(KERN_INFO "DMA-API: preallocated %d debug entries\n",
-			num_entries);
+	pr_info("DMA-API: preallocated %d debug entries\n", num_entries);
 
 	return 0;
 
@@ -349,11 +496,102 @@ out_err:
 	return -ENOMEM;
 }
 
+static ssize_t filter_read(struct file *file, char __user *user_buf,
+			   size_t count, loff_t *ppos)
+{
+	char buf[NAME_MAX_LEN + 1];
+	unsigned long flags;
+	int len;
+
+	if (!current_driver_name[0])
+		return 0;
+
+	/*
+	 * We can't copy to userspace directly because current_driver_name can
+	 * only be read under the driver_name_lock with irqs disabled. So
+	 * create a temporary copy first.
+	 */
+	read_lock_irqsave(&driver_name_lock, flags);
+	len = scnprintf(buf, NAME_MAX_LEN + 1, "%s\n", current_driver_name);
+	read_unlock_irqrestore(&driver_name_lock, flags);
+
+	return simple_read_from_buffer(user_buf, count, ppos, buf, len);
+}
+
+static ssize_t filter_write(struct file *file, const char __user *userbuf,
+			    size_t count, loff_t *ppos)
+{
+	char buf[NAME_MAX_LEN];
+	unsigned long flags;
+	size_t len;
+	int i;
+
+	/*
+	 * We can't copy from userspace directly. Access to
+	 * current_driver_name is protected with a write_lock with irqs
+	 * disabled. Since copy_from_user can fault and may sleep we
+	 * need to copy to temporary buffer first
+	 */
+	len = min(count, (size_t)(NAME_MAX_LEN - 1));
+	if (copy_from_user(buf, userbuf, len))
+		return -EFAULT;
+
+	buf[len] = 0;
+
+	write_lock_irqsave(&driver_name_lock, flags);
+
+	/*
+	 * Now handle the string we got from userspace very carefully.
+	 * The rules are:
+	 *         - only use the first token we got
+	 *         - token delimiter is everything looking like a space
+	 *           character (' ', '\n', '\t' ...)
+	 *
+	 */
+	if (!isalnum(buf[0])) {
+		/*
+		 * If the first character userspace gave us is not
+		 * alphanumerical then assume the filter should be
+		 * switched off.
+		 */
+		if (current_driver_name[0])
+			pr_info("DMA-API: switching off dma-debug driver filter\n");
+		current_driver_name[0] = 0;
+		current_driver = NULL;
+		goto out_unlock;
+	}
+
+	/*
+	 * Now parse out the first token and use it as the name for the
+	 * driver to filter for.
+	 */
+	for (i = 0; i < NAME_MAX_LEN; ++i) {
+		current_driver_name[i] = buf[i];
+		if (isspace(buf[i]) || buf[i] == ' ' || buf[i] == 0)
+			break;
+	}
+	current_driver_name[i] = 0;
+	current_driver = NULL;
+
+	pr_info("DMA-API: enable driver filter for driver [%s]\n",
+		current_driver_name);
+
+out_unlock:
+	write_unlock_irqrestore(&driver_name_lock, flags);
+
+	return count;
+}
+
+const struct file_operations filter_fops = {
+	.read  = filter_read,
+	.write = filter_write,
+};
+
 static int dma_debug_fs_init(void)
 {
 	dma_debug_dent = debugfs_create_dir("dma-api", NULL);
 	if (!dma_debug_dent) {
-		printk(KERN_ERR "DMA-API: can not create debugfs directory\n");
+		pr_err("DMA-API: can not create debugfs directory\n");
 		return -ENOMEM;
 	}
 
@@ -392,6 +630,11 @@ static int dma_debug_fs_init(void)
 	if (!min_free_entries_dent)
 		goto out_err;
 
+	filter_dent = debugfs_create_file("driver_filter", 0644,
+					  dma_debug_dent, NULL, &filter_fops);
+	if (!filter_dent)
+		goto out_err;
+
 	return 0;
 
 out_err:
@@ -400,9 +643,64 @@ out_err:
 	return -ENOMEM;
 }
 
+static int device_dma_allocations(struct device *dev)
+{
+	struct dma_debug_entry *entry;
+	unsigned long flags;
+	int count = 0, i;
+
+	local_irq_save(flags);
+
+	for (i = 0; i < HASH_SIZE; ++i) {
+		spin_lock(&dma_entry_hash[i].lock);
+		list_for_each_entry(entry, &dma_entry_hash[i].list, list) {
+			if (entry->dev == dev)
+				count += 1;
+		}
+		spin_unlock(&dma_entry_hash[i].lock);
+	}
+
+	local_irq_restore(flags);
+
+	return count;
+}
+
+static int dma_debug_device_change(struct notifier_block *nb,
+				    unsigned long action, void *data)
+{
+	struct device *dev = data;
+	int count;
+
+
+	switch (action) {
+	case BUS_NOTIFY_UNBOUND_DRIVER:
+		count = device_dma_allocations(dev);
+		if (count == 0)
+			break;
+		err_printk(dev, NULL, "DMA-API: device driver has pending "
+				"DMA allocations while released from device "
+				"[count=%d]\n", count);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
 void dma_debug_add_bus(struct bus_type *bus)
 {
-	/* FIXME: register notifier */
+	struct notifier_block *nb;
+
+	nb = kzalloc(sizeof(struct notifier_block), GFP_KERNEL);
+	if (nb == NULL) {
+		pr_err("dma_debug_add_bus: out of memory\n");
+		return;
+	}
+
+	nb->notifier_call = dma_debug_device_change;
+
+	bus_register_notifier(bus, nb);
 }
 
 /*
@@ -421,8 +719,7 @@ void dma_debug_init(u32 num_entries)
 	}
 
 	if (dma_debug_fs_init() != 0) {
-		printk(KERN_ERR "DMA-API: error creating debugfs entries "
-				"- disabling\n");
+		pr_err("DMA-API: error creating debugfs entries - disabling\n");
 		global_disable = true;
 
 		return;
@@ -432,14 +729,15 @@ void dma_debug_init(u32 num_entries)
 		num_entries = req_entries;
 
 	if (prealloc_memory(num_entries) != 0) {
-		printk(KERN_ERR "DMA-API: debugging out of memory error "
-				"- disabled\n");
+		pr_err("DMA-API: debugging out of memory error - disabled\n");
 		global_disable = true;
 
 		return;
 	}
 
-	printk(KERN_INFO "DMA-API: debugging enabled by kernel config\n");
+	nr_total_entries = num_free_entries;
+
+	pr_info("DMA-API: debugging enabled by kernel config\n");
 }
 
 static __init int dma_debug_cmdline(char *str)
@@ -448,8 +746,7 @@ static __init int dma_debug_cmdline(char *str)
 		return -EINVAL;
 
 	if (strncmp(str, "off", 3) == 0) {
-		printk(KERN_INFO "DMA-API: debugging disabled on kernel "
-				 "command line\n");
+		pr_info("DMA-API: debugging disabled on kernel command line\n");
 		global_disable = true;
 	}
 
@@ -723,15 +1020,15 @@ void debug_dma_map_sg(struct device *dev, struct scatterlist *sg,
 		entry->type           = dma_debug_sg;
 		entry->dev            = dev;
 		entry->paddr          = sg_phys(s);
-		entry->size           = s->length;
-		entry->dev_addr       = s->dma_address;
+		entry->size           = sg_dma_len(s);
+		entry->dev_addr       = sg_dma_address(s);
 		entry->direction      = direction;
 		entry->sg_call_ents   = nents;
 		entry->sg_mapped_ents = mapped_ents;
 
 		if (!PageHighMem(sg_page(s))) {
 			check_for_stack(dev, sg_virt(s));
-			check_for_illegal_area(dev, sg_virt(s), s->length);
+			check_for_illegal_area(dev, sg_virt(s), sg_dma_len(s));
 		}
 
 		add_dma_entry(entry);
@@ -739,13 +1036,33 @@ void debug_dma_map_sg(struct device *dev, struct scatterlist *sg,
 }
 EXPORT_SYMBOL(debug_dma_map_sg);
 
+static int get_nr_mapped_entries(struct device *dev, struct scatterlist *s)
+{
+	struct dma_debug_entry *entry, ref;
+	struct hash_bucket *bucket;
+	unsigned long flags;
+	int mapped_ents;
+
+	ref.dev      = dev;
+	ref.dev_addr = sg_dma_address(s);
+	ref.size     = sg_dma_len(s),
+
+	bucket       = get_hash_bucket(&ref, &flags);
+	entry        = hash_bucket_find(bucket, &ref);
+	mapped_ents  = 0;
+
+	if (entry)
+		mapped_ents = entry->sg_mapped_ents;
+	put_hash_bucket(bucket, &flags);
+
+	return mapped_ents;
+}
+
 void debug_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
 			int nelems, int dir)
 {
-	struct dma_debug_entry *entry;
 	struct scatterlist *s;
 	int mapped_ents = 0, i;
-	unsigned long flags;
 
 	if (unlikely(global_disable))
 		return;
@@ -756,8 +1073,8 @@ void debug_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
 			.type           = dma_debug_sg,
 			.dev            = dev,
 			.paddr          = sg_phys(s),
-			.dev_addr       = s->dma_address,
-			.size           = s->length,
+			.dev_addr       = sg_dma_address(s),
+			.size           = sg_dma_len(s),
 			.direction      = dir,
 			.sg_call_ents   = 0,
 		};
@@ -765,14 +1082,9 @@ void debug_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
 		if (mapped_ents && i >= mapped_ents)
 			break;
 
-		if (mapped_ents == 0) {
-			struct hash_bucket *bucket;
+		if (!i) {
 			ref.sg_call_ents = nelems;
-			bucket = get_hash_bucket(&ref, &flags);
-			entry = hash_bucket_find(bucket, &ref);
-			if (entry)
-				mapped_ents = entry->sg_mapped_ents;
-			put_hash_bucket(bucket, &flags);
+			mapped_ents = get_nr_mapped_entries(dev, s);
 		}
 
 		check_unmap(&ref);
@@ -874,14 +1186,20 @@ void debug_dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
 			       int nelems, int direction)
 {
 	struct scatterlist *s;
-	int i;
+	int mapped_ents = 0, i;
 
 	if (unlikely(global_disable))
 		return;
 
 	for_each_sg(sg, s, nelems, i) {
-		check_sync(dev, s->dma_address, s->dma_length, 0,
-				direction, true);
+		if (!i)
+			mapped_ents = get_nr_mapped_entries(dev, s);
+
+		if (i >= mapped_ents)
+			break;
+
+		check_sync(dev, sg_dma_address(s), sg_dma_len(s), 0,
+			   direction, true);
 	}
 }
 EXPORT_SYMBOL(debug_dma_sync_sg_for_cpu);
@@ -890,15 +1208,39 @@ void debug_dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
 				  int nelems, int direction)
 {
 	struct scatterlist *s;
-	int i;
+	int mapped_ents = 0, i;
 
 	if (unlikely(global_disable))
 		return;
 
 	for_each_sg(sg, s, nelems, i) {
-		check_sync(dev, s->dma_address, s->dma_length, 0,
-				direction, false);
+		if (!i)
+			mapped_ents = get_nr_mapped_entries(dev, s);
+
+		if (i >= mapped_ents)
+			break;
+
+		check_sync(dev, sg_dma_address(s), sg_dma_len(s), 0,
+			   direction, false);
 	}
 }
 EXPORT_SYMBOL(debug_dma_sync_sg_for_device);
 
+static int __init dma_debug_driver_setup(char *str)
+{
+	int i;
+
+	for (i = 0; i < NAME_MAX_LEN - 1; ++i, ++str) {
+		current_driver_name[i] = *str;
+		if (*str == 0)
+			break;
+	}
+
+	if (current_driver_name[0])
+		pr_info("DMA-API: enable driver filter for driver [%s]\n",
+			current_driver_name);
+
+
+	return 1;
+}
+__setup("dma_debug_driver=", dma_debug_driver_setup);
diff --git a/lib/swiotlb.c b/lib/swiotlb.c
index 2b0b5a7..bffe6d7 100644
--- a/lib/swiotlb.c
+++ b/lib/swiotlb.c
@@ -60,8 +60,8 @@ enum dma_sync_target {
 int swiotlb_force;
 
 /*
- * Used to do a quick range check in swiotlb_unmap_single and
- * swiotlb_sync_single_*, to see if the memory was in fact allocated by this
+ * Used to do a quick range check in unmap_single and
+ * sync_single_*, to see if the memory was in fact allocated by this
  * API.
  */
 static char *io_tlb_start, *io_tlb_end;
@@ -129,7 +129,7 @@ dma_addr_t __weak swiotlb_phys_to_bus(struct device *hwdev, phys_addr_t paddr)
 	return paddr;
 }
 
-phys_addr_t __weak swiotlb_bus_to_phys(dma_addr_t baddr)
+phys_addr_t __weak swiotlb_bus_to_phys(struct device *hwdev, dma_addr_t baddr)
 {
 	return baddr;
 }
@@ -140,9 +140,15 @@ static dma_addr_t swiotlb_virt_to_bus(struct device *hwdev,
 	return swiotlb_phys_to_bus(hwdev, virt_to_phys(address));
 }
 
-static void *swiotlb_bus_to_virt(dma_addr_t address)
+void * __weak swiotlb_bus_to_virt(struct device *hwdev, dma_addr_t address)
 {
-	return phys_to_virt(swiotlb_bus_to_phys(address));
+	return phys_to_virt(swiotlb_bus_to_phys(hwdev, address));
+}
+
+int __weak swiotlb_arch_address_needs_mapping(struct device *hwdev,
+					       dma_addr_t addr, size_t size)
+{
+	return !is_buffer_dma_capable(dma_get_mask(hwdev), addr, size);
 }
 
 int __weak swiotlb_arch_range_needs_mapping(phys_addr_t paddr, size_t size)
@@ -309,10 +315,10 @@ cleanup1:
 	return -ENOMEM;
 }
 
-static int
+static inline int
 address_needs_mapping(struct device *hwdev, dma_addr_t addr, size_t size)
 {
-	return !is_buffer_dma_capable(dma_get_mask(hwdev), addr, size);
+	return swiotlb_arch_address_needs_mapping(hwdev, addr, size);
 }
 
 static inline int range_needs_mapping(phys_addr_t paddr, size_t size)
@@ -341,7 +347,7 @@ static void swiotlb_bounce(phys_addr_t phys, char *dma_addr, size_t size,
 		unsigned long flags;
 
 		while (size) {
-			sz = min(PAGE_SIZE - offset, size);
+			sz = min_t(size_t, PAGE_SIZE - offset, size);
 
 			local_irq_save(flags);
 			buffer = kmap_atomic(pfn_to_page(pfn),
@@ -476,7 +482,7 @@ found:
  * dma_addr is the kernel virtual address of the bounce buffer to unmap.
  */
 static void
-unmap_single(struct device *hwdev, char *dma_addr, size_t size, int dir)
+do_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;
@@ -560,7 +566,6 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 				   size)) {
 		/*
 		 * The allocated memory isn't reachable by the device.
-		 * Fall back on swiotlb_map_single().
 		 */
 		free_pages((unsigned long) ret, order);
 		ret = NULL;
@@ -568,9 +573,8 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 	if (!ret) {
 		/*
 		 * We are either out of memory or the device can't DMA
-		 * to GFP_DMA memory; fall back on
-		 * swiotlb_map_single(), which will grab memory from
-		 * the lowest available address range.
+		 * to GFP_DMA memory; fall back on map_single(), which
+		 * will grab memory from the lowest available address range.
 		 */
 		ret = map_single(hwdev, 0, size, DMA_FROM_DEVICE);
 		if (!ret)
@@ -587,7 +591,7 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 		       (unsigned long long)dev_addr);
 
 		/* DMA_TO_DEVICE to avoid memcpy in unmap_single */
-		unmap_single(hwdev, ret, size, DMA_TO_DEVICE);
+		do_unmap_single(hwdev, ret, size, DMA_TO_DEVICE);
 		return NULL;
 	}
 	*dma_handle = dev_addr;
@@ -604,7 +608,7 @@ swiotlb_free_coherent(struct device *hwdev, size_t size, void *vaddr,
 		free_pages((unsigned long) vaddr, get_order(size));
 	else
 		/* DMA_TO_DEVICE to avoid memcpy in unmap_single */
-		unmap_single(hwdev, vaddr, size, DMA_TO_DEVICE);
+		do_unmap_single(hwdev, vaddr, size, DMA_TO_DEVICE);
 }
 EXPORT_SYMBOL(swiotlb_free_coherent);
 
@@ -634,7 +638,7 @@ swiotlb_full(struct device *dev, size_t size, int dir, int do_panic)
  * physical address to use is returned.
  *
  * Once the device is given the dma address, the device owns this memory until
- * either swiotlb_unmap_single or swiotlb_dma_sync_single is performed.
+ * either swiotlb_unmap_page or swiotlb_dma_sync_single is performed.
  */
 dma_addr_t swiotlb_map_page(struct device *dev, struct page *page,
 			    unsigned long offset, size_t size,
@@ -642,18 +646,17 @@ dma_addr_t swiotlb_map_page(struct device *dev, struct page *page,
 			    struct dma_attrs *attrs)
 {
 	phys_addr_t phys = page_to_phys(page) + offset;
-	void *ptr = page_address(page) + offset;
 	dma_addr_t dev_addr = swiotlb_phys_to_bus(dev, phys);
 	void *map;
 
 	BUG_ON(dir == DMA_NONE);
 	/*
-	 * If the pointer passed in happens to be in the device's DMA window,
+	 * If the address happens to be in the device's DMA window,
 	 * we can safely return the device addr and not worry about bounce
 	 * buffering it.
 	 */
 	if (!address_needs_mapping(dev, dev_addr, size) &&
-	    !range_needs_mapping(virt_to_phys(ptr), size))
+	    !range_needs_mapping(phys, size))
 		return dev_addr;
 
 	/*
@@ -679,23 +682,35 @@ EXPORT_SYMBOL_GPL(swiotlb_map_page);
 
 /*
  * Unmap a single streaming mode DMA translation.  The dma_addr and size must
- * match what was provided for in a previous swiotlb_map_single call.  All
+ * match what was provided for in a previous swiotlb_map_page call.  All
  * other usages are undefined.
  *
  * After this call, reads by the cpu to the buffer are guaranteed to see
  * whatever the device wrote there.
  */
+static void unmap_single(struct device *hwdev, dma_addr_t dev_addr,
+			 size_t size, int dir)
+{
+	char *dma_addr = swiotlb_bus_to_virt(hwdev, dev_addr);
+
+	BUG_ON(dir == DMA_NONE);
+
+	if (is_swiotlb_buffer(dma_addr)) {
+		do_unmap_single(hwdev, dma_addr, size, dir);
+		return;
+	}
+
+	if (dir != DMA_FROM_DEVICE)
+		return;
+
+	dma_mark_clean(dma_addr, size);
+}
+
 void swiotlb_unmap_page(struct device *hwdev, dma_addr_t dev_addr,
 			size_t size, enum dma_data_direction dir,
 			struct dma_attrs *attrs)
 {
-	char *dma_addr = swiotlb_bus_to_virt(dev_addr);
-
-	BUG_ON(dir == DMA_NONE);
-	if (is_swiotlb_buffer(dma_addr))
-		unmap_single(hwdev, dma_addr, size, dir);
-	else if (dir == DMA_FROM_DEVICE)
-		dma_mark_clean(dma_addr, size);
+	unmap_single(hwdev, dev_addr, size, dir);
 }
 EXPORT_SYMBOL_GPL(swiotlb_unmap_page);
 
@@ -703,7 +718,7 @@ EXPORT_SYMBOL_GPL(swiotlb_unmap_page);
  * Make physical memory consistent for a single streaming mode DMA translation
  * after a transfer.
  *
- * If you perform a swiotlb_map_single() but wish to interrogate the buffer
+ * If you perform a swiotlb_map_page() but wish to interrogate the buffer
  * using the cpu, yet do not wish to teardown the dma mapping, you must
  * call this function before doing so.  At the next point you give the dma
  * address back to the card, you must first perform a
@@ -713,13 +728,19 @@ static void
 swiotlb_sync_single(struct device *hwdev, dma_addr_t dev_addr,
 		    size_t size, int dir, int target)
 {
-	char *dma_addr = swiotlb_bus_to_virt(dev_addr);
+	char *dma_addr = swiotlb_bus_to_virt(hwdev, dev_addr);
 
 	BUG_ON(dir == DMA_NONE);
-	if (is_swiotlb_buffer(dma_addr))
+
+	if (is_swiotlb_buffer(dma_addr)) {
 		sync_single(hwdev, dma_addr, size, dir, target);
-	else if (dir == DMA_FROM_DEVICE)
-		dma_mark_clean(dma_addr, size);
+		return;
+	}
+
+	if (dir != DMA_FROM_DEVICE)
+		return;
+
+	dma_mark_clean(dma_addr, size);
 }
 
 void
@@ -746,13 +767,7 @@ 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 = swiotlb_bus_to_virt(dev_addr) + offset;
-
-	BUG_ON(dir == DMA_NONE);
-	if (is_swiotlb_buffer(dma_addr))
-		sync_single(hwdev, dma_addr, size, dir, target);
-	else if (dir == DMA_FROM_DEVICE)
-		dma_mark_clean(dma_addr, size);
+	swiotlb_sync_single(hwdev, dev_addr + offset, size, dir, target);
 }
 
 void
@@ -777,7 +792,7 @@ EXPORT_SYMBOL_GPL(swiotlb_sync_single_range_for_device);
 
 /*
  * Map a set of buffers described by scatterlist in streaming mode for DMA.
- * This is the scatter-gather version of the above swiotlb_map_single
+ * This is the scatter-gather version of the above swiotlb_map_page
  * interface.  Here the scatter gather list elements are each tagged with the
  * appropriate dma address and length.  They are obtained via
  * sg_dma_{address,length}(SG).
@@ -788,7 +803,7 @@ EXPORT_SYMBOL_GPL(swiotlb_sync_single_range_for_device);
  *       The routine returns the number of addr/length pairs actually
  *       used, at most nents.
  *
- * Device ownership issues as mentioned above for swiotlb_map_single are the
+ * Device ownership issues as mentioned above for swiotlb_map_page are the
  * same here.
  */
 int
@@ -836,7 +851,7 @@ EXPORT_SYMBOL(swiotlb_map_sg);
 
 /*
  * Unmap a set of streaming mode DMA translations.  Again, cpu read rules
- * concerning calls here are the same as for swiotlb_unmap_single() above.
+ * concerning calls here are the same as for swiotlb_unmap_page() above.
  */
 void
 swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
@@ -847,13 +862,9 @@ 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 != swiotlb_phys_to_bus(hwdev, sg_phys(sg)))
-			unmap_single(hwdev, swiotlb_bus_to_virt(sg->dma_address),
-				     sg->dma_length, dir);
-		else if (dir == DMA_FROM_DEVICE)
-			dma_mark_clean(swiotlb_bus_to_virt(sg->dma_address), sg->dma_length);
-	}
+	for_each_sg(sgl, sg, nelems, i)
+		unmap_single(hwdev, sg->dma_address, sg->dma_length, dir);
+
 }
 EXPORT_SYMBOL(swiotlb_unmap_sg_attrs);
 
@@ -879,15 +890,9 @@ swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
 	struct scatterlist *sg;
 	int i;
 
-	BUG_ON(dir == DMA_NONE);
-
-	for_each_sg(sgl, sg, nelems, i) {
-		if (sg->dma_address != swiotlb_phys_to_bus(hwdev, sg_phys(sg)))
-			sync_single(hwdev, swiotlb_bus_to_virt(sg->dma_address),
+	for_each_sg(sgl, sg, nelems, i)
+		swiotlb_sync_single(hwdev, sg->dma_address,
 				    sg->dma_length, dir, target);
-		else if (dir == DMA_FROM_DEVICE)
-			dma_mark_clean(swiotlb_bus_to_virt(sg->dma_address), sg->dma_length);
-	}
 }
 
 void