[RFC v2 13/21] RDMA/umem: Prevent UMEM ODP creation with SWIOTLB

Leon Romanovsky leon at kernel.org
Thu Sep 12 04:15:48 PDT 2024


From: Leon Romanovsky <leonro at nvidia.com>

RDMA UMEM never supported DMA addresses returned from SWIOTLB, as these
addresses should be programmed to the hardware which is not aware that
it is bounce buffers and not real ones.

Instead of silently leave broken system for the users who didn't
know it, let's be explicit and return an error to them.

Signed-off-by: Leon Romanovsky <leonro at nvidia.com>
---
 drivers/infiniband/core/umem_odp.c | 78 +++++++++++++++---------------
 drivers/iommu/dma-iommu.c          |  1 +
 2 files changed, 40 insertions(+), 39 deletions(-)

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index 7bfa1e54454c..58fc3d4bfb73 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -42,7 +42,7 @@
 #include <linux/interval_tree.h>
 #include <linux/hmm.h>
 #include <linux/pagemap.h>
-
+#include <linux/iommu-dma.h>
 #include <rdma/ib_umem_odp.h>
 
 #include "uverbs.h"
@@ -51,49 +51,49 @@ static inline int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
 				   const struct mmu_interval_notifier_ops *ops)
 {
 	struct ib_device *dev = umem_odp->umem.ibdev;
+	size_t page_size = 1UL << umem_odp->page_shift;
+	unsigned long start, end;
+	size_t ndmas, npfns;
 	int ret;
 
 	umem_odp->umem.is_odp = 1;
 	mutex_init(&umem_odp->umem_mutex);
+	if (umem_odp->is_implicit_odp)
+		return 0;
+
+	if (!iommu_can_use_iova(dev->dma_device, NULL, page_size,
+				DMA_BIDIRECTIONAL))
+		return -EOPNOTSUPP;
+
+	start = ALIGN_DOWN(umem_odp->umem.address, page_size);
+	if (check_add_overflow(umem_odp->umem.address,
+			       (unsigned long)umem_odp->umem.length, &end))
+		return -EOVERFLOW;
+	end = ALIGN(end, page_size);
+	if (unlikely(end < page_size))
+		return -EOVERFLOW;
+
+	ndmas = (end - start) >> umem_odp->page_shift;
+	if (!ndmas)
+		return -EINVAL;
+
+	npfns = (end - start) >> PAGE_SHIFT;
+	umem_odp->pfn_list =
+		kvcalloc(npfns, sizeof(*umem_odp->pfn_list), GFP_KERNEL);
+	if (!umem_odp->pfn_list)
+		return -ENOMEM;
+
+	dma_init_iova_state(&umem_odp->state, dev->dma_device,
+			    DMA_BIDIRECTIONAL);
+	ret = dma_alloc_iova(&umem_odp->state, end - start);
+	if (ret)
+		goto out_pfn_list;
 
-	if (!umem_odp->is_implicit_odp) {
-		size_t page_size = 1UL << umem_odp->page_shift;
-		unsigned long start;
-		unsigned long end;
-		size_t ndmas, npfns;
-
-		start = ALIGN_DOWN(umem_odp->umem.address, page_size);
-		if (check_add_overflow(umem_odp->umem.address,
-				       (unsigned long)umem_odp->umem.length,
-				       &end))
-			return -EOVERFLOW;
-		end = ALIGN(end, page_size);
-		if (unlikely(end < page_size))
-			return -EOVERFLOW;
-
-		ndmas = (end - start) >> umem_odp->page_shift;
-		if (!ndmas)
-			return -EINVAL;
-
-		npfns = (end - start) >> PAGE_SHIFT;
-		umem_odp->pfn_list = kvcalloc(
-			npfns, sizeof(*umem_odp->pfn_list), GFP_KERNEL);
-		if (!umem_odp->pfn_list)
-			return -ENOMEM;
-
-
-		dma_init_iova_state(&umem_odp->state, dev->dma_device,
-				    DMA_BIDIRECTIONAL);
-		ret = dma_alloc_iova(&umem_odp->state, end - start);
-		if (ret)
-			goto out_pfn_list;
-
-		ret = mmu_interval_notifier_insert(&umem_odp->notifier,
-						   umem_odp->umem.owning_mm,
-						   start, end - start, ops);
-		if (ret)
-			goto out_free_iova;
-	}
+	ret = mmu_interval_notifier_insert(&umem_odp->notifier,
+					   umem_odp->umem.owning_mm, start,
+					   end - start, ops);
+	if (ret)
+		goto out_free_iova;
 
 	return 0;
 
diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
index 3e2e382bb502..af3428ae150d 100644
--- a/drivers/iommu/dma-iommu.c
+++ b/drivers/iommu/dma-iommu.c
@@ -1849,6 +1849,7 @@ bool iommu_can_use_iova(struct device *dev, struct page *page, size_t size,
 
 	return true;
 }
+EXPORT_SYMBOL_GPL(iommu_can_use_iova);
 
 void iommu_setup_dma_ops(struct device *dev)
 {
-- 
2.46.0




More information about the Linux-nvme mailing list