Merge tag 'dma-mapping-6.13-2024-11-19' of git://git.infradead.org/users/hch/dma-mapping

Pull dma-mapping updates from Christoph Hellwig:

 - improve the DMA API tracing code (Sean Anderson)

 - misc cleanups (Christoph Hellwig, Sui Jingfeng)

 - fix pointer abuse when finding the shared DMA pool (Geert
   Uytterhoeven)

 - fix a deadlock in dma-debug (Levi Yun)

* tag 'dma-mapping-6.13-2024-11-19' of git://git.infradead.org/users/hch/dma-mapping:
  dma-mapping: save base/size instead of pointer to shared DMA pool
  dma-mapping: fix swapped dir/flags arguments to trace_dma_alloc_sgt_err
  dma-mapping: drop unneeded includes from dma-mapping.h
  dma-mapping: trace more error paths
  dma-mapping: use trace_dma_alloc for dma_alloc* instead of using trace_dma_map
  dma-mapping: trace dma_alloc/free direction
  dma-mapping: use macros to define events in a class
  dma-mapping: remove an outdated comment from dma-map-ops.h
  dma-debug: remove DMA_API_DEBUG_SG
  dma-debug: store a phys_addr_t in struct dma_debug_entry
  dma-debug: fix a possible deadlock on radix_lock

8 files changed, 242 insertions, 135 deletions

diff --git a/arch/powerpc/platforms/pseries/svm.c b/arch/powerpc/platforms/pseries/svm.c
index 3b4045d508ec..384c9dc1899a 100644
--- a/arch/powerpc/platforms/pseries/svm.c
+++ b/arch/powerpc/platforms/pseries/svm.c
@@ -8,6 +8,7 @@
 
 #include <linux/mm.h>
 #include <linux/memblock.h>
+#include <linux/mem_encrypt.h>
 #include <linux/cc_platform.h>
 #include <asm/machdep.h>
 #include <asm/svm.h>
diff --git a/include/linux/dma-map-ops.h b/include/linux/dma-map-ops.h
index b7773201414c..e172522cd936 100644
--- a/include/linux/dma-map-ops.h
+++ b/include/linux/dma-map-ops.h
@@ -242,7 +242,7 @@ static inline bool dev_is_dma_coherent(struct device *dev)
 {
 	return true;
 }
-#endif /* CONFIG_ARCH_HAS_DMA_COHERENCE_H */
+#endif
 
 static inline void dma_reset_need_sync(struct device *dev)
 {
diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
index 1524da363734..b79925b1c433 100644
--- a/include/linux/dma-mapping.h
+++ b/include/linux/dma-mapping.h
@@ -2,15 +2,11 @@
 #ifndef _LINUX_DMA_MAPPING_H
 #define _LINUX_DMA_MAPPING_H
 
-#include <linux/cache.h>
-#include <linux/sizes.h>
-#include <linux/string.h>
 #include <linux/device.h>
 #include <linux/err.h>
 #include <linux/dma-direction.h>
 #include <linux/scatterlist.h>
 #include <linux/bug.h>
-#include <linux/mem_encrypt.h>
 
 /**
  * List of possible attributes associated with a DMA mapping. The semantics
diff --git a/include/trace/events/dma.h b/include/trace/events/dma.h
index b0f41265191c..d8ddc27b6a7c 100644
--- a/include/trace/events/dma.h
+++ b/include/trace/events/dma.h
@@ -65,15 +65,14 @@ DECLARE_EVENT_CLASS(dma_map,
 		decode_dma_attrs(__entry->attrs))
 );
 
-DEFINE_EVENT(dma_map, dma_map_page,
-	TP_PROTO(struct device *dev, phys_addr_t phys_addr, dma_addr_t dma_addr,
-		 size_t size, enum dma_data_direction dir, unsigned long attrs),
-	TP_ARGS(dev, phys_addr, dma_addr, size, dir, attrs));
+#define DEFINE_MAP_EVENT(name) \
+DEFINE_EVENT(dma_map, name, \
+	TP_PROTO(struct device *dev, phys_addr_t phys_addr, dma_addr_t dma_addr, \
+		 size_t size, enum dma_data_direction dir, unsigned long attrs), \
+	TP_ARGS(dev, phys_addr, dma_addr, size, dir, attrs))
 
-DEFINE_EVENT(dma_map, dma_map_resource,
-	TP_PROTO(struct device *dev, phys_addr_t phys_addr, dma_addr_t dma_addr,
-		 size_t size, enum dma_data_direction dir, unsigned long attrs),
-	TP_ARGS(dev, phys_addr, dma_addr, size, dir, attrs));
+DEFINE_MAP_EVENT(dma_map_page);
+DEFINE_MAP_EVENT(dma_map_resource);
 
 DECLARE_EVENT_CLASS(dma_unmap,
 	TP_PROTO(struct device *dev, dma_addr_t addr, size_t size,
@@ -104,20 +103,20 @@ DECLARE_EVENT_CLASS(dma_unmap,
 		decode_dma_attrs(__entry->attrs))
 );
 
-DEFINE_EVENT(dma_unmap, dma_unmap_page,
-	TP_PROTO(struct device *dev, dma_addr_t addr, size_t size,
-		 enum dma_data_direction dir, unsigned long attrs),
-	TP_ARGS(dev, addr, size, dir, attrs));
+#define DEFINE_UNMAP_EVENT(name) \
+DEFINE_EVENT(dma_unmap, name, \
+	TP_PROTO(struct device *dev, dma_addr_t addr, size_t size, \
+		 enum dma_data_direction dir, unsigned long attrs), \
+	TP_ARGS(dev, addr, size, dir, attrs))
 
-DEFINE_EVENT(dma_unmap, dma_unmap_resource,
-	TP_PROTO(struct device *dev, dma_addr_t addr, size_t size,
-		 enum dma_data_direction dir, unsigned long attrs),
-	TP_ARGS(dev, addr, size, dir, attrs));
+DEFINE_UNMAP_EVENT(dma_unmap_page);
+DEFINE_UNMAP_EVENT(dma_unmap_resource);
 
-TRACE_EVENT(dma_alloc,
+DECLARE_EVENT_CLASS(dma_alloc_class,
 	TP_PROTO(struct device *dev, void *virt_addr, dma_addr_t dma_addr,
-		 size_t size, gfp_t flags, unsigned long attrs),
-	TP_ARGS(dev, virt_addr, dma_addr, size, flags, attrs),
+		 size_t size, enum dma_data_direction dir, gfp_t flags,
+		 unsigned long attrs),
+	TP_ARGS(dev, virt_addr, dma_addr, size, dir, flags, attrs),
 
 	TP_STRUCT__entry(
 		__string(device, dev_name(dev))
@@ -125,6 +124,7 @@ TRACE_EVENT(dma_alloc,
 		__field(u64, dma_addr)
 		__field(size_t, size)
 		__field(gfp_t, flags)
+		__field(enum dma_data_direction, dir)
 		__field(unsigned long, attrs)
 	),
 
@@ -137,8 +137,9 @@ TRACE_EVENT(dma_alloc,
 		__entry->attrs = attrs;
 	),
 
-	TP_printk("%s dma_addr=%llx size=%zu virt_addr=%p flags=%s attrs=%s",
+	TP_printk("%s dir=%s dma_addr=%llx size=%zu virt_addr=%p flags=%s attrs=%s",
 		__get_str(device),
+		decode_dma_data_direction(__entry->dir),
 		__entry->dma_addr,
 		__entry->size,
 		__entry->virt_addr,
@@ -146,16 +147,69 @@ TRACE_EVENT(dma_alloc,
 		decode_dma_attrs(__entry->attrs))
 );
 
-TRACE_EVENT(dma_free,
+#define DEFINE_ALLOC_EVENT(name) \
+DEFINE_EVENT(dma_alloc_class, name, \
+	TP_PROTO(struct device *dev, void *virt_addr, dma_addr_t dma_addr, \
+		 size_t size, enum dma_data_direction dir, gfp_t flags, \
+		 unsigned long attrs), \
+	TP_ARGS(dev, virt_addr, dma_addr, size, dir, flags, attrs))
+
+DEFINE_ALLOC_EVENT(dma_alloc);
+DEFINE_ALLOC_EVENT(dma_alloc_pages);
+DEFINE_ALLOC_EVENT(dma_alloc_sgt_err);
+
+TRACE_EVENT(dma_alloc_sgt,
+	TP_PROTO(struct device *dev, struct sg_table *sgt, size_t size,
+		 enum dma_data_direction dir, gfp_t flags, unsigned long attrs),
+	TP_ARGS(dev, sgt, size, dir, flags, attrs),
+
+	TP_STRUCT__entry(
+		__string(device, dev_name(dev))
+		__dynamic_array(u64, phys_addrs, sgt->orig_nents)
+		__field(u64, dma_addr)
+		__field(size_t, size)
+		__field(enum dma_data_direction, dir)
+		__field(gfp_t, flags)
+		__field(unsigned long, attrs)
+	),
+
+	TP_fast_assign(
+		struct scatterlist *sg;
+		int i;
+
+		__assign_str(device);
+		for_each_sg(sgt->sgl, sg, sgt->orig_nents, i)
+			((u64 *)__get_dynamic_array(phys_addrs))[i] = sg_phys(sg);
+		__entry->dma_addr = sg_dma_address(sgt->sgl);
+		__entry->size = size;
+		__entry->dir = dir;
+		__entry->flags = flags;
+		__entry->attrs = attrs;
+	),
+
+	TP_printk("%s dir=%s dma_addr=%llx size=%zu phys_addrs=%s flags=%s attrs=%s",
+		__get_str(device),
+		decode_dma_data_direction(__entry->dir),
+		__entry->dma_addr,
+		__entry->size,
+		__print_array(__get_dynamic_array(phys_addrs),
+			      __get_dynamic_array_len(phys_addrs) /
+				sizeof(u64), sizeof(u64)),
+		show_gfp_flags(__entry->flags),
+		decode_dma_attrs(__entry->attrs))
+);
+
+DECLARE_EVENT_CLASS(dma_free_class,
 	TP_PROTO(struct device *dev, void *virt_addr, dma_addr_t dma_addr,
-		 size_t size, unsigned long attrs),
-	TP_ARGS(dev, virt_addr, dma_addr, size, attrs),
+		 size_t size, enum dma_data_direction dir, unsigned long attrs),
+	TP_ARGS(dev, virt_addr, dma_addr, size, dir, attrs),
 
 	TP_STRUCT__entry(
 		__string(device, dev_name(dev))
 		__field(void *, virt_addr)
 		__field(u64, dma_addr)
 		__field(size_t, size)
+		__field(enum dma_data_direction, dir)
 		__field(unsigned long, attrs)
 	),
 
@@ -164,17 +218,63 @@ TRACE_EVENT(dma_free,
 		__entry->virt_addr = virt_addr;
 		__entry->dma_addr = dma_addr;
 		__entry->size = size;
+		__entry->dir = dir;
 		__entry->attrs = attrs;
 	),
 
-	TP_printk("%s dma_addr=%llx size=%zu virt_addr=%p attrs=%s",
+	TP_printk("%s dir=%s dma_addr=%llx size=%zu virt_addr=%p attrs=%s",
 		__get_str(device),
+		decode_dma_data_direction(__entry->dir),
 		__entry->dma_addr,
 		__entry->size,
 		__entry->virt_addr,
 		decode_dma_attrs(__entry->attrs))
 );
 
+#define DEFINE_FREE_EVENT(name) \
+DEFINE_EVENT(dma_free_class, name, \
+	TP_PROTO(struct device *dev, void *virt_addr, dma_addr_t dma_addr, \
+		 size_t size, enum dma_data_direction dir, unsigned long attrs), \
+	TP_ARGS(dev, virt_addr, dma_addr, size, dir, attrs))
+
+DEFINE_FREE_EVENT(dma_free);
+DEFINE_FREE_EVENT(dma_free_pages);
+
+TRACE_EVENT(dma_free_sgt,
+	TP_PROTO(struct device *dev, struct sg_table *sgt, size_t size,
+		 enum dma_data_direction dir),
+	TP_ARGS(dev, sgt, size, dir),
+
+	TP_STRUCT__entry(
+		__string(device, dev_name(dev))
+		__dynamic_array(u64, phys_addrs, sgt->orig_nents)
+		__field(u64, dma_addr)
+		__field(size_t, size)
+		__field(enum dma_data_direction, dir)
+	),
+
+	TP_fast_assign(
+		struct scatterlist *sg;
+		int i;
+
+		__assign_str(device);
+		for_each_sg(sgt->sgl, sg, sgt->orig_nents, i)
+			((u64 *)__get_dynamic_array(phys_addrs))[i] = sg_phys(sg);
+		__entry->dma_addr = sg_dma_address(sgt->sgl);
+		__entry->size = size;
+		__entry->dir = dir;
+	),
+
+	TP_printk("%s dir=%s dma_addr=%llx size=%zu phys_addrs=%s",
+		__get_str(device),
+		decode_dma_data_direction(__entry->dir),
+		__entry->dma_addr,
+		__entry->size,
+		__print_array(__get_dynamic_array(phys_addrs),
+			      __get_dynamic_array_len(phys_addrs) /
+				sizeof(u64), sizeof(u64)))
+);
+
 TRACE_EVENT(dma_map_sg,
 	TP_PROTO(struct device *dev, struct scatterlist *sgl, int nents,
 		 int ents, enum dma_data_direction dir, unsigned long attrs),
@@ -221,6 +321,41 @@ TRACE_EVENT(dma_map_sg,
 		decode_dma_attrs(__entry->attrs))
 );
 
+TRACE_EVENT(dma_map_sg_err,
+	TP_PROTO(struct device *dev, struct scatterlist *sgl, int nents,
+		 int err, enum dma_data_direction dir, unsigned long attrs),
+	TP_ARGS(dev, sgl, nents, err, dir, attrs),
+
+	TP_STRUCT__entry(
+		__string(device, dev_name(dev))
+		__dynamic_array(u64, phys_addrs, nents)
+		__field(int, err)
+		__field(enum dma_data_direction, dir)
+		__field(unsigned long, attrs)
+	),
+
+	TP_fast_assign(
+		struct scatterlist *sg;
+		int i;
+
+		__assign_str(device);
+		for_each_sg(sgl, sg, nents, i)
+			((u64 *)__get_dynamic_array(phys_addrs))[i] = sg_phys(sg);
+		__entry->err = err;
+		__entry->dir = dir;
+		__entry->attrs = attrs;
+	),
+
+	TP_printk("%s dir=%s dma_addrs=%s err=%d attrs=%s",
+		__get_str(device),
+		decode_dma_data_direction(__entry->dir),
+		__print_array(__get_dynamic_array(phys_addrs),
+			      __get_dynamic_array_len(phys_addrs) /
+				sizeof(u64), sizeof(u64)),
+		__entry->err,
+		decode_dma_attrs(__entry->attrs))
+);
+
 TRACE_EVENT(dma_unmap_sg,
 	TP_PROTO(struct device *dev, struct scatterlist *sgl, int nents,
 		 enum dma_data_direction dir, unsigned long attrs),
@@ -279,15 +414,14 @@ DECLARE_EVENT_CLASS(dma_sync_single,
 		__entry->size)
 );
 
-DEFINE_EVENT(dma_sync_single, dma_sync_single_for_cpu,
-	TP_PROTO(struct device *dev, dma_addr_t dma_addr, size_t size,
-		 enum dma_data_direction dir),
-	TP_ARGS(dev, dma_addr, size, dir));
+#define DEFINE_SYNC_SINGLE_EVENT(name) \
+DEFINE_EVENT(dma_sync_single, name, \
+	TP_PROTO(struct device *dev, dma_addr_t dma_addr, size_t size, \
+		 enum dma_data_direction dir), \
+	TP_ARGS(dev, dma_addr, size, dir))
 
-DEFINE_EVENT(dma_sync_single, dma_sync_single_for_device,
-	TP_PROTO(struct device *dev, dma_addr_t dma_addr, size_t size,
-		 enum dma_data_direction dir),
-	TP_ARGS(dev, dma_addr, size, dir));
+DEFINE_SYNC_SINGLE_EVENT(dma_sync_single_for_cpu);
+DEFINE_SYNC_SINGLE_EVENT(dma_sync_single_for_device);
 
 DECLARE_EVENT_CLASS(dma_sync_sg,
 	TP_PROTO(struct device *dev, struct scatterlist *sgl, int nents,
@@ -326,15 +460,14 @@ DECLARE_EVENT_CLASS(dma_sync_sg,
 				sizeof(unsigned int), sizeof(unsigned int)))
 );
 
-DEFINE_EVENT(dma_sync_sg, dma_sync_sg_for_cpu,
-	TP_PROTO(struct device *dev, struct scatterlist *sg, int nents,
-		 enum dma_data_direction dir),
-	TP_ARGS(dev, sg, nents, dir));
+#define DEFINE_SYNC_SG_EVENT(name) \
+DEFINE_EVENT(dma_sync_sg, name, \
+	TP_PROTO(struct device *dev, struct scatterlist *sg, int nents, \
+		 enum dma_data_direction dir), \
+	TP_ARGS(dev, sg, nents, dir))
 
-DEFINE_EVENT(dma_sync_sg, dma_sync_sg_for_device,
-	TP_PROTO(struct device *dev, struct scatterlist *sg, int nents,
-		 enum dma_data_direction dir),
-	TP_ARGS(dev, sg, nents, dir));
+DEFINE_SYNC_SG_EVENT(dma_sync_sg_for_cpu);
+DEFINE_SYNC_SG_EVENT(dma_sync_sg_for_device);
 
 #endif /*  _TRACE_DMA_H */
 
diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig
index 4c0dcd909121..31cfdb6b4bc3 100644
--- a/kernel/dma/Kconfig
+++ b/kernel/dma/Kconfig
@@ -260,23 +260,6 @@ config DMA_API_DEBUG
 
 	  If unsure, say N.
 
-config DMA_API_DEBUG_SG
-	bool "Debug DMA scatter-gather usage"
-	default y
-	depends on DMA_API_DEBUG
-	help
-	  Perform extra checking that callers of dma_map_sg() have respected the
-	  appropriate segment length/boundary limits for the given device when
-	  preparing DMA scatterlists.
-
-	  This is particularly likely to have been overlooked in cases where the
-	  dma_map_sg() API is used for general bulk mapping of pages rather than
-	  preparing literal scatter-gather descriptors, where there is a risk of
-	  unexpected behaviour from DMA API implementations if the scatterlist
-	  is technically out-of-spec.
-
-	  If unsure, say N.
-
 config DMA_MAP_BENCHMARK
 	bool "Enable benchmarking of streaming DMA mapping"
 	depends on DEBUG_FS
diff --git a/kernel/dma/coherent.c b/kernel/dma/coherent.c
index ff5683a57f77..3b2bdca9f1d4 100644
--- a/kernel/dma/coherent.c
+++ b/kernel/dma/coherent.c
@@ -330,7 +330,8 @@ int dma_init_global_coherent(phys_addr_t phys_addr, size_t size)
 #include <linux/of_reserved_mem.h>
 
 #ifdef CONFIG_DMA_GLOBAL_POOL
-static struct reserved_mem *dma_reserved_default_memory __initdata;
+static phys_addr_t dma_reserved_default_memory_base __initdata;
+static phys_addr_t dma_reserved_default_memory_size __initdata;
 #endif
 
 static int rmem_dma_device_init(struct reserved_mem *rmem, struct device *dev)
@@ -376,9 +377,10 @@ static int __init rmem_dma_setup(struct reserved_mem *rmem)
 
 #ifdef CONFIG_DMA_GLOBAL_POOL
 	if (of_get_flat_dt_prop(node, "linux,dma-default", NULL)) {
-		WARN(dma_reserved_default_memory,
+		WARN(dma_reserved_default_memory_size,
 		     "Reserved memory: region for default DMA coherent area is redefined\n");
-		dma_reserved_default_memory = rmem;
+		dma_reserved_default_memory_base = rmem->base;
+		dma_reserved_default_memory_size = rmem->size;
 	}
 #endif
 
@@ -391,10 +393,10 @@ static int __init rmem_dma_setup(struct reserved_mem *rmem)
 #ifdef CONFIG_DMA_GLOBAL_POOL
 static int __init dma_init_reserved_memory(void)
 {
-	if (!dma_reserved_default_memory)
+	if (!dma_reserved_default_memory_size)
 		return -ENOMEM;
-	return dma_init_global_coherent(dma_reserved_default_memory->base,
-					dma_reserved_default_memory->size);
+	return dma_init_global_coherent(dma_reserved_default_memory_base,
+					dma_reserved_default_memory_size);
 }
 core_initcall(dma_init_reserved_memory);
 #endif /* CONFIG_DMA_GLOBAL_POOL */
diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c
index d570535342cb..295396226f31 100644
--- a/kernel/dma/debug.c
+++ b/kernel/dma/debug.c
@@ -59,8 +59,7 @@ enum map_err_types {
  * @direction: enum dma_data_direction
  * @sg_call_ents: 'nents' from dma_map_sg
  * @sg_mapped_ents: 'mapped_ents' from dma_map_sg
- * @pfn: page frame of the start address
- * @offset: offset of mapping relative to pfn
+ * @paddr: physical start address of the mapping
  * @map_err_type: track whether dma_mapping_error() was checked
  * @stack_len: number of backtrace entries in @stack_entries
  * @stack_entries: stack of backtrace history
@@ -74,8 +73,7 @@ struct dma_debug_entry {
 	int              direction;
 	int		 sg_call_ents;
 	int		 sg_mapped_ents;
-	unsigned long	 pfn;
-	size_t		 offset;
+	phys_addr_t	 paddr;
 	enum map_err_types  map_err_type;
 #ifdef CONFIG_STACKTRACE
 	unsigned int	stack_len;
@@ -389,14 +387,6 @@ static void hash_bucket_del(struct dma_debug_entry *entry)
 	list_del(&entry->list);
 }
 
-static unsigned long long phys_addr(struct dma_debug_entry *entry)
-{
-	if (entry->type == dma_debug_resource)
-		return __pfn_to_phys(entry->pfn) + entry->offset;
-
-	return page_to_phys(pfn_to_page(entry->pfn)) + entry->offset;
-}
-
 /*
  * For each mapping (initial cacheline in the case of
  * dma_alloc_coherent/dma_map_page, initial cacheline in each page of a
@@ -428,8 +418,8 @@ static DEFINE_SPINLOCK(radix_lock);
 
 static phys_addr_t to_cacheline_number(struct dma_debug_entry *entry)
 {
-	return (entry->pfn << CACHELINE_PER_PAGE_SHIFT) +
-		(entry->offset >> L1_CACHE_SHIFT);
+	return ((entry->paddr >> PAGE_SHIFT) << CACHELINE_PER_PAGE_SHIFT) +
+		(offset_in_page(entry->paddr) >> L1_CACHE_SHIFT);
 }
 
 static int active_cacheline_read_overlap(phys_addr_t cln)
@@ -538,11 +528,11 @@ 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=%llx N=%lx D=%llx L=%llx cln=%pa %s %s\n",
+					 "%s idx %d P=%pa D=%llx L=%llx cln=%pa %s %s\n",
 					 type2name[entry->type], idx,
-					 phys_addr(entry), entry->pfn,
-					 entry->dev_addr, entry->size,
-					 &cln, dir2name[entry->direction],
+					 &entry->paddr, entry->dev_addr,
+					 entry->size, &cln,
+					 dir2name[entry->direction],
 					 maperr2str[entry->map_err_type]);
 			}
 		}
@@ -569,13 +559,13 @@ 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=%llx N=%lx D=%llx L=%llx cln=%pa %s %s\n",
+				   "%s %s %s idx %d P=%pa D=%llx L=%llx cln=%pa %s %s\n",
 				   dev_driver_string(entry->dev),
 				   dev_name(entry->dev),
 				   type2name[entry->type], idx,
-				   phys_addr(entry), entry->pfn,
-				   entry->dev_addr, entry->size,
-				   &cln, dir2name[entry->direction],
+				   &entry->paddr, entry->dev_addr,
+				   entry->size, &cln,
+				   dir2name[entry->direction],
 				   maperr2str[entry->map_err_type]);
 		}
 		spin_unlock_irqrestore(&bucket->lock, flags);
@@ -1003,16 +993,16 @@ static void check_unmap(struct dma_debug_entry *ref)
 			   "[mapped as %s] [unmapped as %s]\n",
 			   ref->dev_addr, ref->size,
 			   type2name[entry->type], type2name[ref->type]);
-	} else if ((entry->type == dma_debug_coherent) &&
-		   (phys_addr(ref) != phys_addr(entry))) {
+	} else if (entry->type == dma_debug_coherent &&
+		   ref->paddr != entry->paddr) {
 		err_printk(ref->dev, entry, "device driver frees "
 			   "DMA memory with different CPU address "
 			   "[device address=0x%016llx] [size=%llu bytes] "
-			   "[cpu alloc address=0x%016llx] "
-			   "[cpu free address=0x%016llx]",
+			   "[cpu alloc address=0x%pa] "
+			   "[cpu free address=0x%pa]",
 			   ref->dev_addr, ref->size,
-			   phys_addr(entry),
-			   phys_addr(ref));
+			   &entry->paddr,
+			   &ref->paddr);
 	}
 
 	if (ref->sg_call_ents && ref->type == dma_debug_sg &&
@@ -1052,9 +1042,13 @@ static void check_unmap(struct dma_debug_entry *ref)
 	}
 
 	hash_bucket_del(entry);
-	dma_entry_free(entry);
-
 	put_hash_bucket(bucket, flags);
+
+	/*
+	 * Free the entry outside of bucket_lock to avoid ABBA deadlocks
+	 * between that and radix_lock.
+	 */
+	dma_entry_free(entry);
 }
 
 static void check_for_stack(struct device *dev,
@@ -1169,7 +1163,6 @@ out:
 
 static void check_sg_segment(struct device *dev, struct scatterlist *sg)
 {
-#ifdef CONFIG_DMA_API_DEBUG_SG
 	unsigned int max_seg = dma_get_max_seg_size(dev);
 	u64 start, end, boundary = dma_get_seg_boundary(dev);
 
@@ -1190,7 +1183,6 @@ static void check_sg_segment(struct device *dev, struct scatterlist *sg)
 	if ((start ^ end) & ~boundary)
 		err_printk(dev, NULL, "mapping sg segment across boundary [start=0x%016llx] [end=0x%016llx] [boundary=0x%016llx]\n",
 			   start, end, boundary);
-#endif
 }
 
 void debug_dma_map_single(struct device *dev, const void *addr,
@@ -1227,8 +1219,7 @@ void debug_dma_map_page(struct device *dev, struct page *page, size_t offset,
 
 	entry->dev       = dev;
 	entry->type      = dma_debug_single;
-	entry->pfn	 = page_to_pfn(page);
-	entry->offset	 = offset;
+	entry->paddr	 = page_to_phys(page);
 	entry->dev_addr  = dma_addr;
 	entry->size      = size;
 	entry->direction = direction;
@@ -1323,8 +1314,7 @@ void debug_dma_map_sg(struct device *dev, struct scatterlist *sg,
 
 		entry->type           = dma_debug_sg;
 		entry->dev            = dev;
-		entry->pfn	      = page_to_pfn(sg_page(s));
-		entry->offset	      = s->offset;
+		entry->paddr	      = sg_phys(s);
 		entry->size           = sg_dma_len(s);
 		entry->dev_addr       = sg_dma_address(s);
 		entry->direction      = direction;
@@ -1370,8 +1360,7 @@ void debug_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
 		struct dma_debug_entry ref = {
 			.type           = dma_debug_sg,
 			.dev            = dev,
-			.pfn		= page_to_pfn(sg_page(s)),
-			.offset		= s->offset,
+			.paddr		= sg_phys(s),
 			.dev_addr       = sg_dma_address(s),
 			.size           = sg_dma_len(s),
 			.direction      = dir,
@@ -1410,16 +1399,12 @@ void debug_dma_alloc_coherent(struct device *dev, size_t size,
 
 	entry->type      = dma_debug_coherent;
 	entry->dev       = dev;
-	entry->offset	 = offset_in_page(virt);
+	entry->paddr	 = page_to_phys((is_vmalloc_addr(virt) ?
+				vmalloc_to_page(virt) : virt_to_page(virt)));
 	entry->size      = size;
 	entry->dev_addr  = dma_addr;
 	entry->direction = DMA_BIDIRECTIONAL;
 
-	if (is_vmalloc_addr(virt))
-		entry->pfn = vmalloc_to_pfn(virt);
-	else
-		entry->pfn = page_to_pfn(virt_to_page(virt));
-
 	add_dma_entry(entry, attrs);
 }
 
@@ -1429,7 +1414,6 @@ void debug_dma_free_coherent(struct device *dev, size_t size,
 	struct dma_debug_entry ref = {
 		.type           = dma_debug_coherent,
 		.dev            = dev,
-		.offset		= offset_in_page(virt),
 		.dev_addr       = dma_addr,
 		.size           = size,
 		.direction      = DMA_BIDIRECTIONAL,
@@ -1439,10 +1423,8 @@ void debug_dma_free_coherent(struct device *dev, size_t size,
 	if (!is_vmalloc_addr(virt) && !virt_addr_valid(virt))
 		return;
 
-	if (is_vmalloc_addr(virt))
-		ref.pfn = vmalloc_to_pfn(virt);
-	else
-		ref.pfn = page_to_pfn(virt_to_page(virt));
+	ref.paddr = page_to_phys((is_vmalloc_addr(virt) ?
+			vmalloc_to_page(virt) : virt_to_page(virt)));
 
 	if (unlikely(dma_debug_disabled()))
 		return;
@@ -1465,8 +1447,7 @@ void debug_dma_map_resource(struct device *dev, phys_addr_t addr, size_t size,
 
 	entry->type		= dma_debug_resource;
 	entry->dev		= dev;
-	entry->pfn		= PHYS_PFN(addr);
-	entry->offset		= offset_in_page(addr);
+	entry->paddr		= addr;
 	entry->size		= size;
 	entry->dev_addr		= dma_addr;
 	entry->direction	= direction;
@@ -1543,8 +1524,7 @@ void debug_dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
 		struct dma_debug_entry ref = {
 			.type           = dma_debug_sg,
 			.dev            = dev,
-			.pfn		= page_to_pfn(sg_page(s)),
-			.offset		= s->offset,
+			.paddr		= sg_phys(s),
 			.dev_addr       = sg_dma_address(s),
 			.size           = sg_dma_len(s),
 			.direction      = direction,
@@ -1575,8 +1555,7 @@ void debug_dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
 		struct dma_debug_entry ref = {
 			.type           = dma_debug_sg,
 			.dev            = dev,
-			.pfn		= page_to_pfn(sg_page(s)),
-			.offset		= s->offset,
+			.paddr		= sg_phys(sg),
 			.dev_addr       = sg_dma_address(s),
 			.size           = sg_dma_len(s),
 			.direction      = direction,
diff --git a/kernel/dma/mapping.c b/kernel/dma/mapping.c
index 864a1121bf08..cda127027e48 100644
--- a/kernel/dma/mapping.c
+++ b/kernel/dma/mapping.c
@@ -223,6 +223,7 @@ static int __dma_map_sg_attrs(struct device *dev, struct scatterlist *sg,
 		debug_dma_map_sg(dev, sg, nents, ents, dir, attrs);
 	} else if (WARN_ON_ONCE(ents != -EINVAL && ents != -ENOMEM &&
 				ents != -EIO && ents != -EREMOTEIO)) {
+		trace_dma_map_sg_err(dev, sg, nents, ents, dir, attrs);
 		return -EIO;
 	}
 
@@ -604,22 +605,29 @@ void *dma_alloc_attrs(struct device *dev, size_t size, dma_addr_t *dma_handle,
 	if (WARN_ON_ONCE(flag & __GFP_COMP))
 		return NULL;
 
-	if (dma_alloc_from_dev_coherent(dev, size, dma_handle, &cpu_addr))
+	if (dma_alloc_from_dev_coherent(dev, size, dma_handle, &cpu_addr)) {
+		trace_dma_alloc(dev, cpu_addr, *dma_handle, size,
+				DMA_BIDIRECTIONAL, flag, attrs);
 		return cpu_addr;
+	}
 
 	/* let the implementation decide on the zone to allocate from: */
 	flag &= ~(__GFP_DMA | __GFP_DMA32 | __GFP_HIGHMEM);
 
-	if (dma_alloc_direct(dev, ops))
+	if (dma_alloc_direct(dev, ops)) {
 		cpu_addr = dma_direct_alloc(dev, size, dma_handle, flag, attrs);
-	else if (use_dma_iommu(dev))
+	} else if (use_dma_iommu(dev)) {
 		cpu_addr = iommu_dma_alloc(dev, size, dma_handle, flag, attrs);
-	else if (ops->alloc)
+	} else if (ops->alloc) {
 		cpu_addr = ops->alloc(dev, size, dma_handle, flag, attrs);
-	else
+	} else {
+		trace_dma_alloc(dev, NULL, 0, size, DMA_BIDIRECTIONAL, flag,
+				attrs);
 		return NULL;
+	}
 
-	trace_dma_alloc(dev, cpu_addr, *dma_handle, size, flag, attrs);
+	trace_dma_alloc(dev, cpu_addr, *dma_handle, size, DMA_BIDIRECTIONAL,
+			flag, attrs);
 	debug_dma_alloc_coherent(dev, size, *dma_handle, cpu_addr, attrs);
 	return cpu_addr;
 }
@@ -641,10 +649,11 @@ void dma_free_attrs(struct device *dev, size_t size, void *cpu_addr,
 	 */
 	WARN_ON(irqs_disabled());
 
+	trace_dma_free(dev, cpu_addr, dma_handle, size, DMA_BIDIRECTIONAL,
+		       attrs);
 	if (!cpu_addr)
 		return;
 
-	trace_dma_free(dev, cpu_addr, dma_handle, size, attrs);
 	debug_dma_free_coherent(dev, size, cpu_addr, dma_handle);
 	if (dma_alloc_direct(dev, ops))
 		dma_direct_free(dev, size, cpu_addr, dma_handle, attrs);
@@ -683,9 +692,11 @@ struct page *dma_alloc_pages(struct device *dev, size_t size,
 	struct page *page = __dma_alloc_pages(dev, size, dma_handle, dir, gfp);
 
 	if (page) {
-		trace_dma_map_page(dev, page_to_phys(page), *dma_handle, size,
-				   dir, 0);
+		trace_dma_alloc_pages(dev, page_to_virt(page), *dma_handle,
+				      size, dir, gfp, 0);
 		debug_dma_map_page(dev, page, 0, size, dir, *dma_handle, 0);
+	} else {
+		trace_dma_alloc_pages(dev, NULL, 0, size, dir, gfp, 0);
 	}
 	return page;
 }
@@ -708,7 +719,7 @@ static void __dma_free_pages(struct device *dev, size_t size, struct page *page,
 void dma_free_pages(struct device *dev, size_t size, struct page *page,
 		dma_addr_t dma_handle, enum dma_data_direction dir)
 {
-	trace_dma_unmap_page(dev, dma_handle, size, dir, 0);
+	trace_dma_free_pages(dev, page_to_virt(page), dma_handle, size, dir, 0);
 	debug_dma_unmap_page(dev, dma_handle, size, dir);
 	__dma_free_pages(dev, size, page, dma_handle, dir);
 }
@@ -768,8 +779,10 @@ struct sg_table *dma_alloc_noncontiguous(struct device *dev, size_t size,
 
 	if (sgt) {
 		sgt->nents = 1;
-		trace_dma_map_sg(dev, sgt->sgl, sgt->orig_nents, 1, dir, attrs);
+		trace_dma_alloc_sgt(dev, sgt, size, dir, gfp, attrs);
 		debug_dma_map_sg(dev, sgt->sgl, sgt->orig_nents, 1, dir, attrs);
+	} else {
+		trace_dma_alloc_sgt_err(dev, NULL, 0, size, dir, gfp, attrs);
 	}
 	return sgt;
 }
@@ -787,7 +800,7 @@ static void free_single_sgt(struct device *dev, size_t size,
 void dma_free_noncontiguous(struct device *dev, size_t size,
 		struct sg_table *sgt, enum dma_data_direction dir)
 {
-	trace_dma_unmap_sg(dev, sgt->sgl, sgt->orig_nents, dir, 0);
+	trace_dma_free_sgt(dev, sgt, size, dir);
 	debug_dma_unmap_sg(dev, sgt->sgl, sgt->orig_nents, dir);
 
 	if (use_dma_iommu(dev))