Merge tag 'dma-mapping-7.2-2026-06-16' of git://git.kernel.org/pub/scm/linux/kernel/git/mszyprowski/linux

Pull dma-mapping updates from Marek Szyprowski:

 - added checks for DMA attributes in the debug code, especially to
   ensure that mappings are created and released with matching
   attributes (Leon Romanovsky)

 - better default configuration for CMA on NUMA machines (Feng Tang)

 - code cleanup in dma benchmark tool (Rosen Penev)

* tag 'dma-mapping-7.2-2026-06-16' of git://git.kernel.org/pub/scm/linux/kernel/git/mszyprowski/linux:
  dma: map_benchmark: turn dma_sg_map_param buf into a flexible array
  dma-contiguous: simplify numa cma area handling
  dma-contiguous: add kconfig option to setup numa cma area if not configured explicitly
  dma-debug: Ensure mappings are created and released with matching attributes
  dma-debug: Feed DMA attribute for unmapping flows too
  dma-debug: Record DMA attributes in debug entry
  dma-debug: Remove unused DMA attribute parameter
  ntb: Use consistent DMA attributes when freeing DMA mappings
  ntb: Store original DMA address for future release

6 files changed, 126 insertions, 95 deletions

diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig
index bfef21b4a9ae..0a4ba21a57a7 100644
--- a/kernel/dma/Kconfig
+++ b/kernel/dma/Kconfig
@@ -179,7 +179,19 @@ config DMA_NUMA_CMA
 
 	  You can set the size of pernuma CMA by specifying "cma_pernuma=size"
 	  or set the node id and its size of CMA by specifying "numa_cma=
-	  <node>:size[,<node>:size]" on the kernel's command line.
+	  <node>:size[,<node>:size]" on the kernel's command line. And in
+	  rare case that the above 2 are both setup, then the "numa_cma="
+	  takes priority for the specified numa nodes.
+
+config CMA_SIZE_PERNUMA
+	bool "Default CMA area per NUMA node"
+	depends on DMA_NUMA_CMA
+	default y
+	help
+	  On systems with more than one NUMA node, the selected CMA
+	  area size will be also allocated on each additional node,
+	  so that most devices may have benefit from better DMA
+	  locality without an explicit command-line opt-in.
 
 comment "Default contiguous memory area size:"
 
diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
index 03f52bd17120..f754079a287d 100644
--- a/kernel/dma/contiguous.c
+++ b/kernel/dma/contiguous.c
@@ -134,8 +134,8 @@ EXPORT_SYMBOL_GPL(dev_get_cma_area);
 
 static struct cma *dma_contiguous_numa_area[MAX_NUMNODES];
 static phys_addr_t numa_cma_size[MAX_NUMNODES] __initdata;
-static struct cma *dma_contiguous_pernuma_area[MAX_NUMNODES];
 static phys_addr_t pernuma_size_bytes __initdata;
+static bool numa_cma_configured __initdata;
 
 static int __init early_numa_cma(char *p)
 {
@@ -164,6 +164,7 @@ static int __init early_numa_cma(char *p)
 			break;
 	}
 
+	numa_cma_configured = true;
 	return 0;
 }
 early_param("numa_cma", early_numa_cma);
@@ -171,6 +172,7 @@ early_param("numa_cma", early_numa_cma);
 static int __init early_cma_pernuma(char *p)
 {
 	pernuma_size_bytes = memparse(p, &p);
+	numa_cma_configured = true;
 	return 0;
 }
 early_param("cma_pernuma", early_cma_pernuma);
@@ -199,8 +201,13 @@ static void __init dma_numa_cma_reserve(void)
 {
 	int nid;
 
+	if (IS_ENABLED(CONFIG_CMA_SIZE_PERNUMA) &&
+	    !numa_cma_configured && dma_contiguous_default_area &&
+	    nr_online_nodes > 1)
+		pernuma_size_bytes = cma_get_size(dma_contiguous_default_area);
+
 	for_each_node(nid) {
-		int ret;
+		int size, ret;
 		char name[CMA_MAX_NAME];
 		struct cma **cma;
 
@@ -210,27 +217,17 @@ static void __init dma_numa_cma_reserve(void)
 			continue;
 		}
 
-		if (pernuma_size_bytes) {
-
-			cma = &dma_contiguous_pernuma_area[nid];
-			snprintf(name, sizeof(name), "pernuma%d", nid);
-			ret = cma_declare_contiguous_nid(0, pernuma_size_bytes, 0, 0,
-							 0, false, name, cma, nid);
-			if (ret)
-				pr_warn("%s: reservation failed: err %d, node %d", __func__,
-					ret, nid);
-		}
-
-		if (numa_cma_size[nid]) {
+		/* per-node numa setting has the priority */
+		size = numa_cma_size[nid] ?: pernuma_size_bytes;
+		if (!size)
+			continue;
 
-			cma = &dma_contiguous_numa_area[nid];
-			snprintf(name, sizeof(name), "numa%d", nid);
-			ret = cma_declare_contiguous_nid(0, numa_cma_size[nid], 0, 0, 0, false,
-							 name, cma, nid);
-			if (ret)
-				pr_warn("%s: reservation failed: err %d, node %d", __func__,
-					ret, nid);
-		}
+		cma = &dma_contiguous_numa_area[nid];
+		snprintf(name, sizeof(name), "numa%d", nid);
+		ret = cma_declare_contiguous_nid(0, size, 0, 0, 0, false, name, cma, nid);
+		if (ret)
+			pr_warn("%s: reservation failed: err %d, node %d", __func__,
+				ret, nid);
 	}
 }
 #else
@@ -255,8 +252,6 @@ void __init dma_contiguous_reserve(phys_addr_t limit)
 	phys_addr_t selected_limit = limit;
 	bool fixed = false;
 
-	dma_numa_cma_reserve();
-
 	pr_debug("%s(limit %08lx)\n", __func__, (unsigned long)limit);
 
 	if (size_cmdline != -1) {
@@ -312,6 +307,8 @@ void __init dma_contiguous_reserve(phys_addr_t limit)
 		if (ret)
 			pr_warn("Couldn't queue default CMA region for heap creation.");
 	}
+
+	dma_numa_cma_reserve();
 }
 
 void __weak
@@ -429,16 +426,8 @@ struct page *dma_alloc_contiguous(struct device *dev, size_t size, gfp_t gfp)
 
 #ifdef CONFIG_DMA_NUMA_CMA
 	if (nid != NUMA_NO_NODE && !(gfp & (GFP_DMA | GFP_DMA32))) {
-		struct cma *cma = dma_contiguous_pernuma_area[nid];
+		struct cma *cma = dma_contiguous_numa_area[nid];
 		struct page *page;
-
-		if (cma) {
-			page = cma_alloc_aligned(cma, size, gfp);
-			if (page)
-				return page;
-		}
-
-		cma = dma_contiguous_numa_area[nid];
 		if (cma) {
 			page = cma_alloc_aligned(cma, size, gfp);
 			if (page)
@@ -476,9 +465,6 @@ void dma_free_contiguous(struct device *dev, struct page *page, size_t size)
 		 * otherwise, page is from either per-numa cma or default cma
 		 */
 #ifdef CONFIG_DMA_NUMA_CMA
-		if (cma_release(dma_contiguous_pernuma_area[page_to_nid(page)],
-					page, count))
-			return;
 		if (cma_release(dma_contiguous_numa_area[page_to_nid(page)],
 					page, count))
 			return;
diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c
index 2c0e2cd89b5e..026d50de340e 100644
--- a/kernel/dma/debug.c
+++ b/kernel/dma/debug.c
@@ -63,7 +63,7 @@ enum map_err_types {
  * @sg_mapped_ents: 'mapped_ents' from dma_map_sg
  * @paddr: physical start address of the mapping
  * @map_err_type: track whether dma_mapping_error() was checked
- * @is_cache_clean: driver promises not to write to buffer while mapped
+ * @attrs: dma attributes
  * @stack_len: number of backtrace entries in @stack_entries
  * @stack_entries: stack of backtrace history
  */
@@ -78,7 +78,7 @@ struct dma_debug_entry {
 	int		 sg_mapped_ents;
 	phys_addr_t	 paddr;
 	enum map_err_types map_err_type;
-	bool		 is_cache_clean;
+	unsigned long	 attrs;
 #ifdef CONFIG_STACKTRACE
 	unsigned int	stack_len;
 	unsigned long	stack_entries[DMA_DEBUG_STACKTRACE_ENTRIES];
@@ -478,6 +478,9 @@ static int active_cacheline_insert(struct dma_debug_entry *entry,
 				   bool *overlap_cache_clean)
 {
 	phys_addr_t cln = to_cacheline_number(entry);
+	bool is_cache_clean = entry->attrs &
+			      (DMA_ATTR_DEBUGGING_IGNORE_CACHELINES |
+			       DMA_ATTR_REQUIRE_COHERENT);
 	unsigned long flags;
 	int rc;
 
@@ -495,12 +498,15 @@ static int active_cacheline_insert(struct dma_debug_entry *entry,
 	if (rc == -EEXIST) {
 		struct dma_debug_entry *existing;
 
-		active_cacheline_inc_overlap(cln, entry->is_cache_clean);
+		active_cacheline_inc_overlap(cln, is_cache_clean);
 		existing = radix_tree_lookup(&dma_active_cacheline, cln);
 		/* A lookup failure here after we got -EEXIST is unexpected. */
 		WARN_ON(!existing);
 		if (existing)
-			*overlap_cache_clean = existing->is_cache_clean;
+			*overlap_cache_clean =
+				existing->attrs &
+				(DMA_ATTR_DEBUGGING_IGNORE_CACHELINES |
+				 DMA_ATTR_REQUIRE_COHERENT);
 	}
 	spin_unlock_irqrestore(&radix_lock, flags);
 
@@ -544,12 +550,13 @@ void debug_dma_dump_mappings(struct device *dev)
 			if (!dev || dev == entry->dev) {
 				cln = to_cacheline_number(entry);
 				dev_info(entry->dev,
-					 "%s idx %d P=%pa D=%llx L=%llx cln=%pa %s %s\n",
+					 "%s idx %d P=%pa D=%llx L=%llx cln=%pa %s %s attrs=0x%lx\n",
 					 type2name[entry->type], idx,
 					 &entry->paddr, entry->dev_addr,
 					 entry->size, &cln,
 					 dir2name[entry->direction],
-					 maperr2str[entry->map_err_type]);
+					 maperr2str[entry->map_err_type],
+					 entry->attrs);
 			}
 		}
 		spin_unlock_irqrestore(&bucket->lock, flags);
@@ -575,14 +582,15 @@ static int dump_show(struct seq_file *seq, void *v)
 		list_for_each_entry(entry, &bucket->list, list) {
 			cln = to_cacheline_number(entry);
 			seq_printf(seq,
-				   "%s %s %s idx %d P=%pa D=%llx L=%llx cln=%pa %s %s\n",
+				   "%s %s %s idx %d P=%pa D=%llx L=%llx cln=%pa %s %s attrs=0x%lx\n",
 				   dev_driver_string(entry->dev),
 				   dev_name(entry->dev),
 				   type2name[entry->type], idx,
 				   &entry->paddr, entry->dev_addr,
 				   entry->size, &cln,
 				   dir2name[entry->direction],
-				   maperr2str[entry->map_err_type]);
+				   maperr2str[entry->map_err_type],
+				   entry->attrs);
 		}
 		spin_unlock_irqrestore(&bucket->lock, flags);
 	}
@@ -594,16 +602,14 @@ DEFINE_SHOW_ATTRIBUTE(dump);
  * Wrapper function for adding an entry to the hash.
  * This function takes care of locking itself.
  */
-static void add_dma_entry(struct dma_debug_entry *entry, unsigned long attrs)
+static void add_dma_entry(struct dma_debug_entry *entry)
 {
+	unsigned long attrs = entry->attrs;
 	bool overlap_cache_clean;
 	struct hash_bucket *bucket;
 	unsigned long flags;
 	int rc;
 
-	entry->is_cache_clean = attrs & (DMA_ATTR_DEBUGGING_IGNORE_CACHELINES |
-					 DMA_ATTR_REQUIRE_COHERENT);
-
 	bucket = get_hash_bucket(entry, &flags);
 	hash_bucket_add(bucket, entry);
 	put_hash_bucket(bucket, flags);
@@ -612,9 +618,10 @@ static void add_dma_entry(struct dma_debug_entry *entry, unsigned long attrs)
 	if (rc == -ENOMEM) {
 		pr_err_once("cacheline tracking ENOMEM, dma-debug disabled\n");
 		global_disable = true;
-	} else if (rc == -EEXIST &&
-		   !(attrs & DMA_ATTR_SKIP_CPU_SYNC) &&
-		   !(entry->is_cache_clean && overlap_cache_clean) &&
+	} else if (rc == -EEXIST && !(attrs & DMA_ATTR_SKIP_CPU_SYNC) &&
+		   !(attrs & (DMA_ATTR_DEBUGGING_IGNORE_CACHELINES |
+			      DMA_ATTR_REQUIRE_COHERENT) &&
+		     overlap_cache_clean) &&
 		   dma_get_cache_alignment() >= L1_CACHE_BYTES &&
 		   !(IS_ENABLED(CONFIG_DMA_BOUNCE_UNALIGNED_KMALLOC) &&
 		     is_swiotlb_active(entry->dev))) {
@@ -1067,6 +1074,29 @@ static void check_unmap(struct dma_debug_entry *ref)
 			   type2name[entry->type]);
 	}
 
+	/*
+	 * This may be no bug in reality - but DMA API still expects
+	 * that entry is unmapped with same attributes as it was mapped.
+	 *
+	 * DMA_ATTR_UNMAP_VALID lists the attributes that must be identical
+	 * between map and unmap. Any attribute outside this set (e.g.
+	 * DMA_ATTR_NO_WARN, DMA_ATTR_SKIP_CPU_SYNC) is allowed to differ.
+	 */
+#define DMA_ATTR_UNMAP_VALID                                               \
+	(DMA_ATTR_NO_KERNEL_MAPPING | DMA_ATTR_FORCE_CONTIGUOUS |          \
+	 DMA_ATTR_MMIO | DMA_ATTR_REQUIRE_COHERENT | DMA_ATTR_PRIVILEGED | \
+	 DMA_ATTR_CC_SHARED)
+	if ((ref->attrs & DMA_ATTR_UNMAP_VALID) !=
+	    (entry->attrs & DMA_ATTR_UNMAP_VALID)) {
+		err_printk(ref->dev, entry,
+			   "device driver frees "
+			   "DMA memory with different attributes "
+			   "[device address=0x%016llx] [size=%llu bytes] "
+			   "[mapped with 0x%lx] [unmapped with 0x%lx]\n",
+			   ref->dev_addr, ref->size, entry->attrs, ref->attrs);
+	}
+#undef DMA_ATTR_UNMAP_VALID
+
 	hash_bucket_del(entry);
 	put_hash_bucket(bucket, flags);
 
@@ -1250,6 +1280,7 @@ void debug_dma_map_phys(struct device *dev, phys_addr_t phys, size_t size,
 	entry->size      = size;
 	entry->direction = direction;
 	entry->map_err_type = MAP_ERR_NOT_CHECKED;
+	entry->attrs     = attrs;
 
 	if (attrs & DMA_ATTR_MMIO) {
 		unsigned long pfn = PHYS_PFN(phys);
@@ -1265,7 +1296,7 @@ void debug_dma_map_phys(struct device *dev, phys_addr_t phys, size_t size,
 			check_for_illegal_area(dev, phys_to_virt(phys), size);
 	}
 
-	add_dma_entry(entry, attrs);
+	add_dma_entry(entry);
 }
 
 void debug_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
@@ -1306,8 +1337,8 @@ void debug_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
 }
 EXPORT_SYMBOL(debug_dma_mapping_error);
 
-void debug_dma_unmap_phys(struct device *dev, dma_addr_t dma_addr,
-			  size_t size, int direction)
+void debug_dma_unmap_phys(struct device *dev, dma_addr_t dma_addr, size_t size,
+			  int direction, unsigned long attrs)
 {
 	struct dma_debug_entry ref = {
 		.type           = dma_debug_phy,
@@ -1315,6 +1346,7 @@ void debug_dma_unmap_phys(struct device *dev, dma_addr_t dma_addr,
 		.dev_addr       = dma_addr,
 		.size           = size,
 		.direction      = direction,
+		.attrs          = attrs,
 	};
 
 	if (unlikely(dma_debug_disabled()))
@@ -1352,10 +1384,11 @@ void debug_dma_map_sg(struct device *dev, struct scatterlist *sg,
 		entry->direction      = direction;
 		entry->sg_call_ents   = nents;
 		entry->sg_mapped_ents = mapped_ents;
+		entry->attrs          = attrs;
 
 		check_sg_segment(dev, s);
 
-		add_dma_entry(entry, attrs);
+		add_dma_entry(entry);
 	}
 }
 
@@ -1379,7 +1412,7 @@ static int get_nr_mapped_entries(struct device *dev,
 }
 
 void debug_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
-			int nelems, int dir)
+			int nelems, int dir, unsigned long attrs)
 {
 	struct scatterlist *s;
 	int mapped_ents = 0, i;
@@ -1397,6 +1430,7 @@ void debug_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
 			.size           = sg_dma_len(s),
 			.direction      = dir,
 			.sg_call_ents   = nelems,
+			.attrs          = attrs,
 		};
 
 		if (mapped_ents && i >= mapped_ents)
@@ -1447,12 +1481,13 @@ void debug_dma_alloc_coherent(struct device *dev, size_t size,
 	entry->size      = size;
 	entry->dev_addr  = dma_addr;
 	entry->direction = DMA_BIDIRECTIONAL;
+	entry->attrs     = attrs;
 
-	add_dma_entry(entry, attrs);
+	add_dma_entry(entry);
 }
 
-void debug_dma_free_coherent(struct device *dev, size_t size,
-			 void *virt, dma_addr_t dma_addr)
+void debug_dma_free_coherent(struct device *dev, size_t size, void *virt,
+			     dma_addr_t dma_addr, unsigned long attrs)
 {
 	struct dma_debug_entry ref = {
 		.type           = dma_debug_coherent,
@@ -1460,6 +1495,7 @@ void debug_dma_free_coherent(struct device *dev, size_t size,
 		.dev_addr       = dma_addr,
 		.size           = size,
 		.direction      = DMA_BIDIRECTIONAL,
+		.attrs          = attrs,
 	};
 
 	/* handle vmalloc and linear addresses */
@@ -1574,8 +1610,7 @@ void debug_dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
 
 void debug_dma_alloc_pages(struct device *dev, struct page *page,
 			   size_t size, int direction,
-			   dma_addr_t dma_addr,
-			   unsigned long attrs)
+			   dma_addr_t dma_addr)
 {
 	struct dma_debug_entry *entry;
 
@@ -1593,7 +1628,7 @@ void debug_dma_alloc_pages(struct device *dev, struct page *page,
 	entry->dev_addr  = dma_addr;
 	entry->direction = direction;
 
-	add_dma_entry(entry, attrs);
+	add_dma_entry(entry);
 }
 
 void debug_dma_free_pages(struct device *dev, struct page *page,
diff --git a/kernel/dma/debug.h b/kernel/dma/debug.h
index da7be0bddcf6..13e384633c32 100644
--- a/kernel/dma/debug.h
+++ b/kernel/dma/debug.h
@@ -14,21 +14,22 @@ extern void debug_dma_map_phys(struct device *dev, phys_addr_t phys,
 			       unsigned long attrs);
 
 extern void debug_dma_unmap_phys(struct device *dev, dma_addr_t addr,
-				 size_t size, int direction);
+				 size_t size, int direction,
+				 unsigned long attrs);
 
 extern void debug_dma_map_sg(struct device *dev, struct scatterlist *sg,
 			     int nents, int mapped_ents, int direction,
 			     unsigned long attrs);
 
 extern void debug_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
-			       int nelems, int dir);
+			       int nelems, int dir, unsigned long attrs);
 
 extern void debug_dma_alloc_coherent(struct device *dev, size_t size,
 				     dma_addr_t dma_addr, void *virt,
 				     unsigned long attrs);
 
-extern void debug_dma_free_coherent(struct device *dev, size_t size,
-				    void *virt, dma_addr_t addr);
+extern void debug_dma_free_coherent(struct device *dev, size_t size, void *virt,
+				    dma_addr_t addr, unsigned long attrs);
 
 extern void debug_dma_sync_single_for_cpu(struct device *dev,
 					  dma_addr_t dma_handle, size_t size,
@@ -47,8 +48,7 @@ extern void debug_dma_sync_sg_for_device(struct device *dev,
 					 int nelems, int direction);
 extern void debug_dma_alloc_pages(struct device *dev, struct page *page,
 				  size_t size, int direction,
-				  dma_addr_t dma_addr,
-				  unsigned long attrs);
+				  dma_addr_t dma_addr);
 extern void debug_dma_free_pages(struct device *dev, struct page *page,
 				 size_t size, int direction,
 				 dma_addr_t dma_addr);
@@ -60,7 +60,8 @@ static inline void debug_dma_map_phys(struct device *dev, phys_addr_t phys,
 }
 
 static inline void debug_dma_unmap_phys(struct device *dev, dma_addr_t addr,
-					size_t size, int direction)
+					size_t size, int direction,
+					unsigned long attrs)
 {
 }
 
@@ -71,8 +72,8 @@ static inline void debug_dma_map_sg(struct device *dev, struct scatterlist *sg,
 }
 
 static inline void debug_dma_unmap_sg(struct device *dev,
-				      struct scatterlist *sglist,
-				      int nelems, int dir)
+				      struct scatterlist *sglist, int nelems,
+				      int dir, unsigned long attrs)
 {
 }
 
@@ -83,7 +84,8 @@ static inline void debug_dma_alloc_coherent(struct device *dev, size_t size,
 }
 
 static inline void debug_dma_free_coherent(struct device *dev, size_t size,
-					   void *virt, dma_addr_t addr)
+					   void *virt, dma_addr_t addr,
+					   unsigned long attrs)
 {
 }
 
@@ -113,8 +115,7 @@ static inline void debug_dma_sync_sg_for_device(struct device *dev,
 
 static inline void debug_dma_alloc_pages(struct device *dev, struct page *page,
 					 size_t size, int direction,
-					 dma_addr_t dma_addr,
-					 unsigned long attrs)
+					 dma_addr_t dma_addr)
 {
 }
 
diff --git a/kernel/dma/map_benchmark.c b/kernel/dma/map_benchmark.c
index 29eeb5fdf199..fdc070f419f6 100644
--- a/kernel/dma/map_benchmark.c
+++ b/kernel/dma/map_benchmark.c
@@ -121,35 +121,35 @@ static struct map_benchmark_ops dma_single_map_benchmark_ops = {
 struct dma_sg_map_param {
 	struct sg_table sgt;
 	struct device *dev;
-	void **buf;
 	u32 npages;
 	u32 dma_dir;
+	void *buf[] __counted_by(npages);
 };
 
 static void *dma_sg_map_benchmark_prepare(struct map_benchmark_data *map)
 {
+	struct dma_sg_map_param *params;
 	struct scatterlist *sg;
+	u32 npages;
 	int i;
 
-	struct dma_sg_map_param *params = kzalloc(sizeof(*params), GFP_KERNEL);
-
-	if (!params)
-		return NULL;
 	/*
 	 * Set the number of scatterlist entries based on the granule.
 	 * In SG mode, 'granule' represents the number of scatterlist entries.
 	 * Each scatterlist entry corresponds to a single page.
 	 */
-	params->npages = map->bparam.granule;
+	npages = map->bparam.granule;
+
+	params = kzalloc_flex(*params, buf, npages);
+	if (!params)
+		return NULL;
+
+	params->npages = npages;
 	params->dma_dir = map->bparam.dma_dir;
 	params->dev = map->dev;
-	params->buf = kmalloc_array(params->npages, sizeof(*params->buf),
-				    GFP_KERNEL);
-	if (!params->buf)
-		goto out;
 
 	if (sg_alloc_table(&params->sgt, params->npages, GFP_KERNEL))
-		goto free_buf;
+		goto free_params;
 
 	for_each_sgtable_sg(&params->sgt, sg, i) {
 		params->buf[i] = (void *)__get_free_page(GFP_KERNEL);
@@ -166,9 +166,7 @@ free_page:
 		free_page((unsigned long)params->buf[i]);
 
 	sg_free_table(&params->sgt);
-free_buf:
-	kfree(params->buf);
-out:
+free_params:
 	kfree(params);
 	return NULL;
 }
@@ -183,7 +181,6 @@ static void dma_sg_map_benchmark_unprepare(void *mparam)
 
 	sg_free_table(&params->sgt);
 
-	kfree(params->buf);
 	kfree(params);
 }
 
diff --git a/kernel/dma/mapping.c b/kernel/dma/mapping.c
index 4eedb1a6273a..4fe04669e5e6 100644
--- a/kernel/dma/mapping.c
+++ b/kernel/dma/mapping.c
@@ -223,7 +223,7 @@ void dma_unmap_phys(struct device *dev, dma_addr_t addr, size_t size,
 	else if (ops->unmap_phys)
 		ops->unmap_phys(dev, addr, size, dir, attrs);
 	trace_dma_unmap_phys(dev, addr, size, dir, attrs);
-	debug_dma_unmap_phys(dev, addr, size, dir);
+	debug_dma_unmap_phys(dev, addr, size, dir, attrs);
 }
 EXPORT_SYMBOL_GPL(dma_unmap_phys);
 
@@ -349,7 +349,7 @@ void dma_unmap_sg_attrs(struct device *dev, struct scatterlist *sg,
 
 	BUG_ON(!valid_dma_direction(dir));
 	trace_dma_unmap_sg(dev, sg, nents, dir, attrs);
-	debug_dma_unmap_sg(dev, sg, nents, dir);
+	debug_dma_unmap_sg(dev, sg, nents, dir, attrs);
 	if (dma_map_direct(dev, ops) ||
 	    arch_dma_unmap_sg_direct(dev, sg, nents))
 		dma_direct_unmap_sg(dev, sg, nents, dir, attrs);
@@ -687,7 +687,7 @@ void dma_free_attrs(struct device *dev, size_t size, void *cpu_addr,
 	if (!cpu_addr)
 		return;
 
-	debug_dma_free_coherent(dev, size, cpu_addr, dma_handle);
+	debug_dma_free_coherent(dev, size, cpu_addr, dma_handle, attrs);
 	if (dma_alloc_direct(dev, ops) || arch_dma_free_direct(dev, dma_handle))
 		dma_direct_free(dev, size, cpu_addr, dma_handle, attrs);
 	else if (use_dma_iommu(dev))
@@ -727,7 +727,7 @@ struct page *dma_alloc_pages(struct device *dev, size_t size,
 	if (page) {
 		trace_dma_alloc_pages(dev, page_to_virt(page), *dma_handle,
 				      size, dir, gfp, 0);
-		debug_dma_alloc_pages(dev, page, size, dir, *dma_handle, 0);
+		debug_dma_alloc_pages(dev, page, size, dir, *dma_handle);
 	} else {
 		trace_dma_alloc_pages(dev, NULL, 0, size, dir, gfp, 0);
 	}
@@ -834,7 +834,7 @@ void dma_free_noncontiguous(struct device *dev, size_t size,
 		struct sg_table *sgt, enum dma_data_direction dir)
 {
 	trace_dma_free_sgt(dev, sgt, size, dir);
-	debug_dma_unmap_sg(dev, sgt->sgl, sgt->orig_nents, dir);
+	debug_dma_unmap_sg(dev, sgt->sgl, sgt->orig_nents, dir, 0);
 
 	if (use_dma_iommu(dev))
 		iommu_dma_free_noncontiguous(dev, size, sgt, dir);