Merge tag 'dma-mapping-6.11-2024-07-19' of git://git.infradead.org/users/hch/dma-mapping

Pull dma-mapping updates from Christoph Hellwig:

 - reduce duplicate swiotlb pool lookups (Michael Kelley)

 - minor small fixes (Yicong Yang, Yang Li)

* tag 'dma-mapping-6.11-2024-07-19' of git://git.infradead.org/users/hch/dma-mapping:
  swiotlb: fix kernel-doc description for swiotlb_del_transient
  swiotlb: reduce swiotlb pool lookups
  dma-mapping: benchmark: Don't starve others when doing the test

drivers/iommu/dma-iommu.c

@@ -1078,8 +1078,7 @@ static void iommu_dma_sync_single_for_cpu(struct device *dev,
 	if (!dev_is_dma_coherent(dev))
 		arch_sync_dma_for_cpu(phys, size, dir);
 
-	if (is_swiotlb_buffer(dev, phys))
-		swiotlb_sync_single_for_cpu(dev, phys, size, dir);
+	swiotlb_sync_single_for_cpu(dev, phys, size, dir);
 }
 
 static void iommu_dma_sync_single_for_device(struct device *dev,
@@ -1091,8 +1090,7 @@ static void iommu_dma_sync_single_for_device(struct device *dev,
 		return;
 
 	phys = iommu_iova_to_phys(iommu_get_dma_domain(dev), dma_handle);
-	if (is_swiotlb_buffer(dev, phys))
-		swiotlb_sync_single_for_device(dev, phys, size, dir);
+	swiotlb_sync_single_for_device(dev, phys, size, dir);
 
 	if (!dev_is_dma_coherent(dev))
 		arch_sync_dma_for_device(phys, size, dir);
@@ -1186,7 +1184,7 @@ static dma_addr_t iommu_dma_map_page(struct device *dev, struct page *page,
 		arch_sync_dma_for_device(phys, size, dir);
 
 	iova = __iommu_dma_map(dev, phys, size, prot, dma_mask);
-	if (iova == DMA_MAPPING_ERROR && is_swiotlb_buffer(dev, phys))
+	if (iova == DMA_MAPPING_ERROR)
 		swiotlb_tbl_unmap_single(dev, phys, size, dir, attrs);
 	return iova;
 }
@@ -1206,8 +1204,7 @@ static void iommu_dma_unmap_page(struct device *dev, dma_addr_t dma_handle,
 
 	__iommu_dma_unmap(dev, dma_handle, size);
 
-	if (unlikely(is_swiotlb_buffer(dev, phys)))
-		swiotlb_tbl_unmap_single(dev, phys, size, dir, attrs);
+	swiotlb_tbl_unmap_single(dev, phys, size, dir, attrs);
 }
 
 /*

drivers/xen/swiotlb-xen.c

@@ -88,7 +88,8 @@ static inline int range_straddles_page_boundary(phys_addr_t p, size_t size)
 	return 0;
 }
 
-static int is_xen_swiotlb_buffer(struct device *dev, dma_addr_t dma_addr)
+static struct io_tlb_pool *xen_swiotlb_find_pool(struct device *dev,
+						 dma_addr_t dma_addr)
 {
 	unsigned long bfn = XEN_PFN_DOWN(dma_to_phys(dev, dma_addr));
 	unsigned long xen_pfn = bfn_to_local_pfn(bfn);
@@ -99,8 +100,8 @@ static int is_xen_swiotlb_buffer(struct device *dev, dma_addr_t dma_addr)
 	 * in our domain. Therefore _only_ check address within our domain.
 	 */
 	if (pfn_valid(PFN_DOWN(paddr)))
-		return is_swiotlb_buffer(dev, paddr);
-	return 0;
+		return swiotlb_find_pool(dev, paddr);
+	return NULL;
 }
 
 #ifdef CONFIG_X86
@@ -227,8 +228,9 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
 	 * Ensure that the address returned is DMA'ble
 	 */
 	if (unlikely(!dma_capable(dev, dev_addr, size, true))) {
-		swiotlb_tbl_unmap_single(dev, map, size, dir,
-				attrs | DMA_ATTR_SKIP_CPU_SYNC);
+		__swiotlb_tbl_unmap_single(dev, map, size, dir,
+				attrs | DMA_ATTR_SKIP_CPU_SYNC,
+				swiotlb_find_pool(dev, map));
 		return DMA_MAPPING_ERROR;
 	}
 
@@ -254,6 +256,7 @@ static void xen_swiotlb_unmap_page(struct device *hwdev, dma_addr_t dev_addr,
 		size_t size, enum dma_data_direction dir, unsigned long attrs)
 {
 	phys_addr_t paddr = xen_dma_to_phys(hwdev, dev_addr);
+	struct io_tlb_pool *pool;
 
 	BUG_ON(dir == DMA_NONE);
 
@@ -265,8 +268,10 @@ static void xen_swiotlb_unmap_page(struct device *hwdev, dma_addr_t dev_addr,
 	}
 
 	/* NOTE: We use dev_addr here, not paddr! */
-	if (is_xen_swiotlb_buffer(hwdev, dev_addr))
-		swiotlb_tbl_unmap_single(hwdev, paddr, size, dir, attrs);
+	pool = xen_swiotlb_find_pool(hwdev, dev_addr);
+	if (pool)
+		__swiotlb_tbl_unmap_single(hwdev, paddr, size, dir,
+					   attrs, pool);
 }
 
 static void
@@ -274,6 +279,7 @@ xen_swiotlb_sync_single_for_cpu(struct device *dev, dma_addr_t dma_addr,
 		size_t size, enum dma_data_direction dir)
 {
 	phys_addr_t paddr = xen_dma_to_phys(dev, dma_addr);
+	struct io_tlb_pool *pool;
 
 	if (!dev_is_dma_coherent(dev)) {
 		if (pfn_valid(PFN_DOWN(dma_to_phys(dev, dma_addr))))
@@ -282,8 +288,9 @@ xen_swiotlb_sync_single_for_cpu(struct device *dev, dma_addr_t dma_addr,
 			xen_dma_sync_for_cpu(dev, dma_addr, size, dir);
 	}
 
-	if (is_xen_swiotlb_buffer(dev, dma_addr))
-		swiotlb_sync_single_for_cpu(dev, paddr, size, dir);
+	pool = xen_swiotlb_find_pool(dev, dma_addr);
+	if (pool)
+		__swiotlb_sync_single_for_cpu(dev, paddr, size, dir, pool);
 }
 
 static void
@@ -291,9 +298,11 @@ xen_swiotlb_sync_single_for_device(struct device *dev, dma_addr_t dma_addr,
 		size_t size, enum dma_data_direction dir)
 {
 	phys_addr_t paddr = xen_dma_to_phys(dev, dma_addr);
+	struct io_tlb_pool *pool;
 
-	if (is_xen_swiotlb_buffer(dev, dma_addr))
-		swiotlb_sync_single_for_device(dev, paddr, size, dir);
+	pool = xen_swiotlb_find_pool(dev, dma_addr);
+	if (pool)
+		__swiotlb_sync_single_for_device(dev, paddr, size, dir, pool);
 
 	if (!dev_is_dma_coherent(dev)) {
 		if (pfn_valid(PFN_DOWN(dma_to_phys(dev, dma_addr))))

include/linux/scatterlist.h

@@ -332,7 +332,7 @@ static inline void sg_dma_unmark_bus_address(struct scatterlist *sg)
  * Description:
  *   Returns true if the scatterlist was marked for SWIOTLB bouncing. Not all
  *   elements may have been bounced, so the caller would have to check
- *   individual SG entries with is_swiotlb_buffer().
+ *   individual SG entries with swiotlb_find_pool().
  */
 static inline bool sg_dma_is_swiotlb(struct scatterlist *sg)
 {

include/linux/swiotlb.h

@@ -42,24 +42,6 @@ int swiotlb_init_late(size_t size, gfp_t gfp_mask,
 	int (*remap)(void *tlb, unsigned long nslabs));
 extern void __init swiotlb_update_mem_attributes(void);
 
-phys_addr_t swiotlb_tbl_map_single(struct device *hwdev, phys_addr_t phys,
-		size_t mapping_size,
-		unsigned int alloc_aligned_mask, enum dma_data_direction dir,
-		unsigned long attrs);
-
-extern void swiotlb_tbl_unmap_single(struct device *hwdev,
-				     phys_addr_t tlb_addr,
-				     size_t mapping_size,
-				     enum dma_data_direction dir,
-				     unsigned long attrs);
-
-void swiotlb_sync_single_for_device(struct device *dev, phys_addr_t tlb_addr,
-		size_t size, enum dma_data_direction dir);
-void swiotlb_sync_single_for_cpu(struct device *dev, phys_addr_t tlb_addr,
-		size_t size, enum dma_data_direction dir);
-dma_addr_t swiotlb_map(struct device *dev, phys_addr_t phys,
-		size_t size, enum dma_data_direction dir, unsigned long attrs);
-
 #ifdef CONFIG_SWIOTLB
 
 /**
@@ -143,37 +125,27 @@ struct io_tlb_mem {
 #endif
 };
 
-#ifdef CONFIG_SWIOTLB_DYNAMIC
-
-struct io_tlb_pool *swiotlb_find_pool(struct device *dev, phys_addr_t paddr);
-
-#else
-
-static inline struct io_tlb_pool *swiotlb_find_pool(struct device *dev,
-						    phys_addr_t paddr)
-{
-	return &dev->dma_io_tlb_mem->defpool;
-}
-
-#endif
+struct io_tlb_pool *__swiotlb_find_pool(struct device *dev, phys_addr_t paddr);
 
 /**
- * is_swiotlb_buffer() - check if a physical address belongs to a swiotlb
+ * swiotlb_find_pool() - find swiotlb pool to which a physical address belongs
  * @dev:        Device which has mapped the buffer.
  * @paddr:      Physical address within the DMA buffer.
  *
- * Check if @paddr points into a bounce buffer.
+ * Find the swiotlb pool that @paddr points into.
  *
  * Return:
- * * %true if @paddr points into a bounce buffer
- * * %false otherwise
+ * * pool address if @paddr points into a bounce buffer
+ * * NULL if @paddr does not point into a bounce buffer. As such, this function
+ *   can be used to determine if @paddr denotes a swiotlb bounce buffer.
  */
-static inline bool is_swiotlb_buffer(struct device *dev, phys_addr_t paddr)
+static inline struct io_tlb_pool *swiotlb_find_pool(struct device *dev,
+		phys_addr_t paddr)
 {
 	struct io_tlb_mem *mem = dev->dma_io_tlb_mem;
 
 	if (!mem)
-		return false;
+		return NULL;
 
 #ifdef CONFIG_SWIOTLB_DYNAMIC
 	/*
@@ -182,16 +154,19 @@ static inline bool is_swiotlb_buffer(struct device *dev, phys_addr_t paddr)
 	 * If a SWIOTLB address is checked on another CPU, then it was
 	 * presumably loaded by the device driver from an unspecified private
 	 * data structure. Make sure that this load is ordered before reading
-	 * dev->dma_uses_io_tlb here and mem->pools in swiotlb_find_pool().
+	 * dev->dma_uses_io_tlb here and mem->pools in __swiotlb_find_pool().
 	 *
 	 * This barrier pairs with smp_mb() in swiotlb_find_slots().
 	 */
 	smp_rmb();
-	return READ_ONCE(dev->dma_uses_io_tlb) &&
-		swiotlb_find_pool(dev, paddr);
+	if (READ_ONCE(dev->dma_uses_io_tlb))
+		return __swiotlb_find_pool(dev, paddr);
 #else
-	return paddr >= mem->defpool.start && paddr < mem->defpool.end;
+	if (paddr >= mem->defpool.start && paddr < mem->defpool.end)
+		return &mem->defpool;
 #endif
+
+	return NULL;
 }
 
 static inline bool is_swiotlb_force_bounce(struct device *dev)
@@ -219,9 +194,10 @@ static inline void swiotlb_dev_init(struct device *dev)
 {
 }
 
-static inline bool is_swiotlb_buffer(struct device *dev, phys_addr_t paddr)
+static inline struct io_tlb_pool *swiotlb_find_pool(struct device *dev,
+		phys_addr_t paddr)
 {
-	return false;
+	return NULL;
 }
 static inline bool is_swiotlb_force_bounce(struct device *dev)
 {
@@ -260,6 +236,49 @@ static inline phys_addr_t default_swiotlb_limit(void)
 }
 #endif /* CONFIG_SWIOTLB */
 
+phys_addr_t swiotlb_tbl_map_single(struct device *hwdev, phys_addr_t phys,
+		size_t mapping_size, unsigned int alloc_aligned_mask,
+		enum dma_data_direction dir, unsigned long attrs);
+dma_addr_t swiotlb_map(struct device *dev, phys_addr_t phys,
+		size_t size, enum dma_data_direction dir, unsigned long attrs);
+
+void __swiotlb_tbl_unmap_single(struct device *hwdev, phys_addr_t tlb_addr,
+		size_t mapping_size, enum dma_data_direction dir,
+		unsigned long attrs, struct io_tlb_pool *pool);
+static inline void swiotlb_tbl_unmap_single(struct device *dev,
+		phys_addr_t addr, size_t size, enum dma_data_direction dir,
+		unsigned long attrs)
+{
+	struct io_tlb_pool *pool = swiotlb_find_pool(dev, addr);
+
+	if (unlikely(pool))
+		__swiotlb_tbl_unmap_single(dev, addr, size, dir, attrs, pool);
+}
+
+void __swiotlb_sync_single_for_device(struct device *dev, phys_addr_t tlb_addr,
+		size_t size, enum dma_data_direction dir,
+		struct io_tlb_pool *pool);
+static inline void swiotlb_sync_single_for_device(struct device *dev,
+		phys_addr_t addr, size_t size, enum dma_data_direction dir)
+{
+	struct io_tlb_pool *pool = swiotlb_find_pool(dev, addr);
+
+	if (unlikely(pool))
+		__swiotlb_sync_single_for_device(dev, addr, size, dir, pool);
+}
+
+void __swiotlb_sync_single_for_cpu(struct device *dev, phys_addr_t tlb_addr,
+		size_t size, enum dma_data_direction dir,
+		struct io_tlb_pool *pool);
+static inline void swiotlb_sync_single_for_cpu(struct device *dev,
+		phys_addr_t addr, size_t size, enum dma_data_direction dir)
+{
+	struct io_tlb_pool *pool = swiotlb_find_pool(dev, addr);
+
+	if (unlikely(pool))
+		__swiotlb_sync_single_for_cpu(dev, addr, size, dir, pool);
+}
+
 extern void swiotlb_print_info(void);
 
 #ifdef CONFIG_DMA_RESTRICTED_POOL

kernel/dma/direct.c

@@ -404,9 +404,7 @@ void dma_direct_sync_sg_for_device(struct device *dev,
 	for_each_sg(sgl, sg, nents, i) {
 		phys_addr_t paddr = dma_to_phys(dev, sg_dma_address(sg));
 
-		if (unlikely(is_swiotlb_buffer(dev, paddr)))
-			swiotlb_sync_single_for_device(dev, paddr, sg->length,
-						       dir);
+		swiotlb_sync_single_for_device(dev, paddr, sg->length, dir);
 
 		if (!dev_is_dma_coherent(dev))
 			arch_sync_dma_for_device(paddr, sg->length,
@@ -430,9 +428,7 @@ void dma_direct_sync_sg_for_cpu(struct device *dev,
 		if (!dev_is_dma_coherent(dev))
 			arch_sync_dma_for_cpu(paddr, sg->length, dir);
 
-		if (unlikely(is_swiotlb_buffer(dev, paddr)))
-			swiotlb_sync_single_for_cpu(dev, paddr, sg->length,
-						    dir);
+		swiotlb_sync_single_for_cpu(dev, paddr, sg->length, dir);
 
 		if (dir == DMA_FROM_DEVICE)
 			arch_dma_mark_clean(paddr, sg->length);
@@ -640,7 +636,7 @@ size_t dma_direct_max_mapping_size(struct device *dev)
 bool dma_direct_need_sync(struct device *dev, dma_addr_t dma_addr)
 {
 	return !dev_is_dma_coherent(dev) ||
-	       is_swiotlb_buffer(dev, dma_to_phys(dev, dma_addr));
+	       swiotlb_find_pool(dev, dma_to_phys(dev, dma_addr));
 }
 
 /**

kernel/dma/direct.h

@@ -58,8 +58,7 @@ static inline void dma_direct_sync_single_for_device(struct device *dev,
 {
 	phys_addr_t paddr = dma_to_phys(dev, addr);
 
-	if (unlikely(is_swiotlb_buffer(dev, paddr)))
-		swiotlb_sync_single_for_device(dev, paddr, size, dir);
+	swiotlb_sync_single_for_device(dev, paddr, size, dir);
 
 	if (!dev_is_dma_coherent(dev))
 		arch_sync_dma_for_device(paddr, size, dir);
@@ -75,8 +74,7 @@ static inline void dma_direct_sync_single_for_cpu(struct device *dev,
 		arch_sync_dma_for_cpu_all();
 	}
 
-	if (unlikely(is_swiotlb_buffer(dev, paddr)))
-		swiotlb_sync_single_for_cpu(dev, paddr, size, dir);
+	swiotlb_sync_single_for_cpu(dev, paddr, size, dir);
 
 	if (dir == DMA_FROM_DEVICE)
 		arch_dma_mark_clean(paddr, size);
@@ -121,8 +119,7 @@ static inline void dma_direct_unmap_page(struct device *dev, dma_addr_t addr,
 	if (!(attrs & DMA_ATTR_SKIP_CPU_SYNC))
 		dma_direct_sync_single_for_cpu(dev, addr, size, dir);
 
-	if (unlikely(is_swiotlb_buffer(dev, phys)))
-		swiotlb_tbl_unmap_single(dev, phys, size, dir,
+	swiotlb_tbl_unmap_single(dev, phys, size, dir,
 					 attrs | DMA_ATTR_SKIP_CPU_SYNC);
 }
 #endif /* _KERNEL_DMA_DIRECT_H */

kernel/dma/map_benchmark.c

@@ -89,6 +89,22 @@ static int map_benchmark_thread(void *data)
 		atomic64_add(map_sq, &map->sum_sq_map);
 		atomic64_add(unmap_sq, &map->sum_sq_unmap);
 		atomic64_inc(&map->loops);
+
+		/*
+		 * We may test for a long time so periodically check whether
+		 * we need to schedule to avoid starving the others. Otherwise
+		 * we may hangup the kernel in a non-preemptible kernel when
+		 * the test kthreads number >= CPU number, the test kthreads
+		 * will run endless on every CPU since the thread resposible
+		 * for notifying the kthread stop (in do_map_benchmark())
+		 * could not be scheduled.
+		 *
+		 * Note this may degrade the test concurrency since the test
+		 * threads may need to share the CPU time with other load
+		 * in the system. So it's recommended to run this benchmark
+		 * on an idle system.
+		 */
+		cond_resched();
 	}
 
 out:

kernel/dma/swiotlb.c

@@ -763,16 +763,18 @@ static void swiotlb_dyn_free(struct rcu_head *rcu)
 }
 
 /**
- * swiotlb_find_pool() - find the IO TLB pool for a physical address
+ * __swiotlb_find_pool() - find the IO TLB pool for a physical address
  * @dev:        Device which has mapped the DMA buffer.
  * @paddr:      Physical address within the DMA buffer.
  *
  * Find the IO TLB memory pool descriptor which contains the given physical
- * address, if any.
+ * address, if any. This function is for use only when the dev is known to
+ * be using swiotlb. Use swiotlb_find_pool() for the more general case
+ * when this condition is not met.
  *
  * Return: Memory pool which contains @paddr, or %NULL if none.
  */
-struct io_tlb_pool *swiotlb_find_pool(struct device *dev, phys_addr_t paddr)
+struct io_tlb_pool *__swiotlb_find_pool(struct device *dev, phys_addr_t paddr)
 {
 	struct io_tlb_mem *mem = dev->dma_io_tlb_mem;
 	struct io_tlb_pool *pool;
@@ -855,9 +857,8 @@ static unsigned int swiotlb_align_offset(struct device *dev,
  * Bounce: copy the swiotlb buffer from or back to the original dma location
  */
 static void swiotlb_bounce(struct device *dev, phys_addr_t tlb_addr, size_t size,
-			   enum dma_data_direction dir)
+			   enum dma_data_direction dir, struct io_tlb_pool *mem)
 {
-	struct io_tlb_pool *mem = swiotlb_find_pool(dev, tlb_addr);
 	int index = (tlb_addr - mem->start) >> IO_TLB_SHIFT;
 	phys_addr_t orig_addr = mem->slots[index].orig_addr;
 	size_t alloc_size = mem->slots[index].alloc_size;
@@ -1243,7 +1244,7 @@ static int swiotlb_find_slots(struct device *dev, phys_addr_t orig_addr,
 	 * that was made by swiotlb_dyn_alloc() on a third CPU (cf. multicopy
 	 * atomicity).
 	 *
-	 * See also the comment in is_swiotlb_buffer().
+	 * See also the comment in swiotlb_find_pool().
 	 */
 	smp_mb();
 
@@ -1435,13 +1436,13 @@ phys_addr_t swiotlb_tbl_map_single(struct device *dev, phys_addr_t orig_addr,
 	 * hardware behavior.  Use of swiotlb is supposed to be transparent,
 	 * i.e. swiotlb must not corrupt memory by clobbering unwritten bytes.
 	 */
-	swiotlb_bounce(dev, tlb_addr, mapping_size, DMA_TO_DEVICE);
+	swiotlb_bounce(dev, tlb_addr, mapping_size, DMA_TO_DEVICE, pool);
 	return tlb_addr;
 }
 
-static void swiotlb_release_slots(struct device *dev, phys_addr_t tlb_addr)
+static void swiotlb_release_slots(struct device *dev, phys_addr_t tlb_addr,
+				  struct io_tlb_pool *mem)
 {
-	struct io_tlb_pool *mem = swiotlb_find_pool(dev, tlb_addr);
 	unsigned long flags;
 	unsigned int offset = swiotlb_align_offset(dev, 0, tlb_addr);
 	int index, nslots, aindex;
@@ -1499,17 +1500,16 @@ static void swiotlb_release_slots(struct device *dev, phys_addr_t tlb_addr)
  * swiotlb_del_transient() - delete a transient memory pool
  * @dev:	Device which mapped the buffer.
  * @tlb_addr:	Physical address within a bounce buffer.
+ * @pool:       Pointer to the transient memory pool to be checked and deleted.
  *
  * Check whether the address belongs to a transient SWIOTLB memory pool.
  * If yes, then delete the pool.
  *
  * Return: %true if @tlb_addr belonged to a transient pool that was released.
  */
-static bool swiotlb_del_transient(struct device *dev, phys_addr_t tlb_addr)
+static bool swiotlb_del_transient(struct device *dev, phys_addr_t tlb_addr,
+		struct io_tlb_pool *pool)
 {
-	struct io_tlb_pool *pool;
-
-	pool = swiotlb_find_pool(dev, tlb_addr);
 	if (!pool->transient)
 		return false;
 
@@ -1522,7 +1522,7 @@ static bool swiotlb_del_transient(struct device *dev, phys_addr_t tlb_addr)
 #else  /* !CONFIG_SWIOTLB_DYNAMIC */
 
 static inline bool swiotlb_del_transient(struct device *dev,
-					 phys_addr_t tlb_addr)
+		phys_addr_t tlb_addr, struct io_tlb_pool *pool)
 {
 	return false;
 }
@@ -1532,36 +1532,39 @@ static inline bool swiotlb_del_transient(struct device *dev,
 /*
  * tlb_addr is the physical address of the bounce buffer to unmap.
  */
-void swiotlb_tbl_unmap_single(struct device *dev, phys_addr_t tlb_addr,
-			      size_t mapping_size, enum dma_data_direction dir,
-			      unsigned long attrs)
+void __swiotlb_tbl_unmap_single(struct device *dev, phys_addr_t tlb_addr,
+		size_t mapping_size, enum dma_data_direction dir,
+		unsigned long attrs, struct io_tlb_pool *pool)
 {
 	/*
 	 * First, sync the memory before unmapping the entry
 	 */
 	if (!(attrs & DMA_ATTR_SKIP_CPU_SYNC) &&
 	    (dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL))
-		swiotlb_bounce(dev, tlb_addr, mapping_size, DMA_FROM_DEVICE);
+		swiotlb_bounce(dev, tlb_addr, mapping_size,
+						DMA_FROM_DEVICE, pool);
 
-	if (swiotlb_del_transient(dev, tlb_addr))
+	if (swiotlb_del_transient(dev, tlb_addr, pool))
 		return;
-	swiotlb_release_slots(dev, tlb_addr);
+	swiotlb_release_slots(dev, tlb_addr, pool);
 }
 
-void swiotlb_sync_single_for_device(struct device *dev, phys_addr_t tlb_addr,
-		size_t size, enum dma_data_direction dir)
+void __swiotlb_sync_single_for_device(struct device *dev, phys_addr_t tlb_addr,
+		size_t size, enum dma_data_direction dir,
+		struct io_tlb_pool *pool)
 {
 	if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)
-		swiotlb_bounce(dev, tlb_addr, size, DMA_TO_DEVICE);
+		swiotlb_bounce(dev, tlb_addr, size, DMA_TO_DEVICE, pool);
 	else
 		BUG_ON(dir != DMA_FROM_DEVICE);
 }
 
-void swiotlb_sync_single_for_cpu(struct device *dev, phys_addr_t tlb_addr,
-		size_t size, enum dma_data_direction dir)
+void __swiotlb_sync_single_for_cpu(struct device *dev, phys_addr_t tlb_addr,
+		size_t size, enum dma_data_direction dir,
+		struct io_tlb_pool *pool)
 {
 	if (dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL)
-		swiotlb_bounce(dev, tlb_addr, size, DMA_FROM_DEVICE);
+		swiotlb_bounce(dev, tlb_addr, size, DMA_FROM_DEVICE, pool);
 	else
 		BUG_ON(dir != DMA_TO_DEVICE);
 }
@@ -1585,8 +1588,9 @@ dma_addr_t swiotlb_map(struct device *dev, phys_addr_t paddr, size_t size,
 	/* Ensure that the address returned is DMA'ble */
 	dma_addr = phys_to_dma_unencrypted(dev, swiotlb_addr);
 	if (unlikely(!dma_capable(dev, dma_addr, size, true))) {
-		swiotlb_tbl_unmap_single(dev, swiotlb_addr, size, dir,
-			attrs | DMA_ATTR_SKIP_CPU_SYNC);
+		__swiotlb_tbl_unmap_single(dev, swiotlb_addr, size, dir,
+			attrs | DMA_ATTR_SKIP_CPU_SYNC,
+			swiotlb_find_pool(dev, swiotlb_addr));
 		dev_WARN_ONCE(dev, 1,
 			"swiotlb addr %pad+%zu overflow (mask %llx, bus limit %llx).\n",
 			&dma_addr, size, *dev->dma_mask, dev->bus_dma_limit);
@@ -1764,7 +1768,7 @@ struct page *swiotlb_alloc(struct device *dev, size_t size)
 	if (unlikely(!PAGE_ALIGNED(tlb_addr))) {
 		dev_WARN_ONCE(dev, 1, "Cannot allocate pages from non page-aligned swiotlb addr 0x%pa.\n",
 			      &tlb_addr);
-		swiotlb_release_slots(dev, tlb_addr);
+		swiotlb_release_slots(dev, tlb_addr, pool);
 		return NULL;
 	}
 
@@ -1774,11 +1778,13 @@ struct page *swiotlb_alloc(struct device *dev, size_t size)
 bool swiotlb_free(struct device *dev, struct page *page, size_t size)
 {
 	phys_addr_t tlb_addr = page_to_phys(page);
+	struct io_tlb_pool *pool;
 
-	if (!is_swiotlb_buffer(dev, tlb_addr))
+	pool = swiotlb_find_pool(dev, tlb_addr);
+	if (!pool)
 		return false;
 
-	swiotlb_release_slots(dev, tlb_addr);
+	swiotlb_release_slots(dev, tlb_addr, pool);
 
 	return true;
 }