[PATCH 0/2] vhost: fix physical address mapping

[PATCH 0/2] vhost: fix physical address mapping

[xuan.ding]

From: Xuan Ding <xuan.ding@intel.com>

This patchset fixes the issue of incorrect DMA mapping in PA mode. 
Due to the ambiguity of host_phys_addr naming in the guest page
struct, rename it to host_iova.

Xuan Ding (2):
  vhost: fix physical address mapping
  vhost: rename field in guest page struct

 lib/vhost/vhost.h      |  11 ++--
 lib/vhost/vhost_user.c | 130 ++++++++++++++++++++---------------------
 lib/vhost/virtio_net.c |  11 ++--
 3 files changed, 75 insertions(+), 77 deletions(-)

-- 
2.17.1

[PATCH 1/2] vhost: fix physical address mapping

[xuan.ding]

From: Xuan Ding <xuan.ding@intel.com>

When choosing IOVA as PA mode, IOVA is likely to be discontinuous,
which requires page by page mapping for DMA devices. To be consistent,
this patch implements page by page mapping instead of mapping at the
region granularity for both IOVA as VA and PA mode.

Fixes: 7c61fa08b716 ("vhost: enable IOMMU for async vhost")

Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Signed-off-by: Yuan Wang <yuanx.wang@intel.com>
---
 lib/vhost/vhost.h      |   1 +
 lib/vhost/vhost_user.c | 116 ++++++++++++++++++++---------------------
 2 files changed, 57 insertions(+), 60 deletions(-)

diff --git a/lib/vhost/vhost.h b/lib/vhost/vhost.h
index 7085e0885c..d246538ca5 100644
--- a/lib/vhost/vhost.h
+++ b/lib/vhost/vhost.h
@@ -355,6 +355,7 @@ struct vring_packed_desc_event {
 struct guest_page {
 	uint64_t guest_phys_addr;
 	uint64_t host_phys_addr;
+	uint64_t host_user_addr;
 	uint64_t size;
 };
 
diff --git a/lib/vhost/vhost_user.c b/lib/vhost/vhost_user.c
index a781346c4d..6d888766b0 100644
--- a/lib/vhost/vhost_user.c
+++ b/lib/vhost/vhost_user.c
@@ -143,57 +143,56 @@ get_blk_size(int fd)
 	return ret == -1 ? (uint64_t)-1 : (uint64_t)stat.st_blksize;
 }
 
-static int
-async_dma_map(struct rte_vhost_mem_region *region, bool do_map)
+static void
+async_dma_map(struct virtio_net *dev, bool do_map)
 {
-	uint64_t host_iova;
 	int ret = 0;
-
-	host_iova = rte_mem_virt2iova((void *)(uintptr_t)region->host_user_addr);
+	uint32_t i;
+	struct guest_page *page;
 	if (do_map) {
-		/* Add mapped region into the default container of DPDK. */
-		ret = rte_vfio_container_dma_map(RTE_VFIO_DEFAULT_CONTAINER_FD,
-						 region->host_user_addr,
-						 host_iova,
-						 region->size);
-		if (ret) {
-			/*
-			 * DMA device may bind with kernel driver, in this case,
-			 * we don't need to program IOMMU manually. However, if no
-			 * device is bound with vfio/uio in DPDK, and vfio kernel
-			 * module is loaded, the API will still be called and return
-			 * with ENODEV/ENOSUP.
-			 *
-			 * DPDK vfio only returns ENODEV/ENOSUP in very similar
-			 * situations(vfio either unsupported, or supported
-			 * but no devices found). Either way, no mappings could be
-			 * performed. We treat it as normal case in async path.
-			 */
-			if (rte_errno == ENODEV || rte_errno == ENOTSUP)
-				return 0;
-
-			VHOST_LOG_CONFIG(ERR, "DMA engine map failed\n");
-			/* DMA mapping errors won't stop VHST_USER_SET_MEM_TABLE. */
-			return 0;
+		for (i = 0; i < dev->nr_guest_pages; i++) {
+			page = &dev->guest_pages[i];
+			ret = rte_vfio_container_dma_map(RTE_VFIO_DEFAULT_CONTAINER_FD,
+							 page->host_user_addr,
+							 page->host_phys_addr,
+							 page->size);
+			if (ret) {
+				/*
+				 * DMA device may bind with kernel driver, in this case,
+				 * we don't need to program IOMMU manually. However, if no
+				 * device is bound with vfio/uio in DPDK, and vfio kernel
+				 * module is loaded, the API will still be called and return
+				 * with ENODEV.
+				 *
+				 * DPDK vfio only returns ENODEV in very similar situations
+				 * (vfio either unsupported, or supported but no devices found).
+				 * Either way, no mappings could be performed. We treat it as
+				 * normal case in async path. This is a workaround.
+				 */
+				if (rte_errno == ENODEV)
+					return;
+
+				/* DMA mapping errors won't stop VHOST_USER_SET_MEM_TABLE. */
+				VHOST_LOG_CONFIG(ERR, "DMA engine map failed\n");
+			}
 		}
 
 	} else {
-		/* Remove mapped region from the default container of DPDK. */
-		ret = rte_vfio_container_dma_unmap(RTE_VFIO_DEFAULT_CONTAINER_FD,
-						   region->host_user_addr,
-						   host_iova,
-						   region->size);
-		if (ret) {
-			/* like DMA map, ignore the kernel driver case when unmap. */
-			if (rte_errno == EINVAL)
-				return 0;
-
-			VHOST_LOG_CONFIG(ERR, "DMA engine unmap failed\n");
-			return ret;
+		for (i = 0; i < dev->nr_guest_pages; i++) {
+			page = &dev->guest_pages[i];
+			ret = rte_vfio_container_dma_unmap(RTE_VFIO_DEFAULT_CONTAINER_FD,
+							   page->host_user_addr,
+							   page->host_phys_addr,
+							   page->size);
+			if (ret) {
+				/* like DMA map, ignore the kernel driver case when unmap. */
+				if (rte_errno == EINVAL)
+					return;
+
+				VHOST_LOG_CONFIG(ERR, "DMA engine unmap failed\n");
+			}
 		}
 	}
-
-	return ret;
 }
 
 static void
@@ -205,12 +204,12 @@ free_mem_region(struct virtio_net *dev)
 	if (!dev || !dev->mem)
 		return;
 
+	if (dev->async_copy && rte_vfio_is_enabled("vfio"))
+		async_dma_map(dev, false);
+
 	for (i = 0; i < dev->mem->nregions; i++) {
 		reg = &dev->mem->regions[i];
 		if (reg->host_user_addr) {
-			if (dev->async_copy && rte_vfio_is_enabled("vfio"))
-				async_dma_map(reg, false);
-
 			munmap(reg->mmap_addr, reg->mmap_size);
 			close(reg->fd);
 		}
@@ -978,7 +977,7 @@ vhost_user_set_vring_base(struct virtio_net **pdev,
 
 static int
 add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
-		   uint64_t host_phys_addr, uint64_t size)
+		   uint64_t host_phys_addr, uint64_t host_user_addr, uint64_t size)
 {
 	struct guest_page *page, *last_page;
 	struct guest_page *old_pages;
@@ -999,8 +998,9 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 	if (dev->nr_guest_pages > 0) {
 		last_page = &dev->guest_pages[dev->nr_guest_pages - 1];
 		/* merge if the two pages are continuous */
-		if (host_phys_addr == last_page->host_phys_addr +
-				      last_page->size) {
+		if (host_phys_addr == last_page->host_phys_addr + last_page->size
+		    && guest_phys_addr == last_page->guest_phys_addr + last_page->size
+		    && host_user_addr == last_page->host_user_addr + last_page->size) {
 			last_page->size += size;
 			return 0;
 		}
@@ -1009,6 +1009,7 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 	page = &dev->guest_pages[dev->nr_guest_pages++];
 	page->guest_phys_addr = guest_phys_addr;
 	page->host_phys_addr  = host_phys_addr;
+	page->host_user_addr = host_user_addr;
 	page->size = size;
 
 	return 0;
@@ -1028,7 +1029,8 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 	size = page_size - (guest_phys_addr & (page_size - 1));
 	size = RTE_MIN(size, reg_size);
 
-	if (add_one_guest_page(dev, guest_phys_addr, host_phys_addr, size) < 0)
+	if (add_one_guest_page(dev, guest_phys_addr, host_phys_addr,
+			       host_user_addr, size) < 0)
 		return -1;
 
 	host_user_addr  += size;
@@ -1040,7 +1042,7 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 		host_phys_addr = rte_mem_virt2iova((void *)(uintptr_t)
 						  host_user_addr);
 		if (add_one_guest_page(dev, guest_phys_addr, host_phys_addr,
-				size) < 0)
+				       host_user_addr, size) < 0)
 			return -1;
 
 		host_user_addr  += size;
@@ -1215,7 +1217,6 @@ vhost_user_mmap_region(struct virtio_net *dev,
 	uint64_t mmap_size;
 	uint64_t alignment;
 	int populate;
-	int ret;
 
 	/* Check for memory_size + mmap_offset overflow */
 	if (mmap_offset >= -region->size) {
@@ -1274,14 +1275,6 @@ vhost_user_mmap_region(struct virtio_net *dev,
 			VHOST_LOG_CONFIG(ERR, "adding guest pages to region failed.\n");
 			return -1;
 		}
-
-		if (rte_vfio_is_enabled("vfio")) {
-			ret = async_dma_map(region, true);
-			if (ret) {
-				VHOST_LOG_CONFIG(ERR, "Configure IOMMU for DMA engine failed\n");
-				return -1;
-			}
-		}
 	}
 
 	VHOST_LOG_CONFIG(INFO,
@@ -1420,6 +1413,9 @@ vhost_user_set_mem_table(struct virtio_net **pdev, struct VhostUserMsg *msg,
 		dev->mem->nregions++;
 	}
 
+	if (dev->async_copy && rte_vfio_is_enabled("vfio"))
+		async_dma_map(dev, true);
+
 	if (vhost_user_postcopy_register(dev, main_fd, msg) < 0)
 		goto free_mem_table;
 
-- 
2.17.1

[PATCH 2/2] vhost: rename field in guest page struct

[xuan.ding]

From: Xuan Ding <xuan.ding@intel.com>

This patch renames the host_phys_addr to host_iova in guest_page
struct. The host_phys_addr is iova, it depends on the DPDK
IOVA mode.

Signed-off-by: Xuan Ding <xuan.ding@intel.com>
---
 lib/vhost/vhost.h      | 10 +++++-----
 lib/vhost/vhost_user.c | 24 ++++++++++++------------
 lib/vhost/virtio_net.c | 11 ++++++-----
 3 files changed, 23 insertions(+), 22 deletions(-)

diff --git a/lib/vhost/vhost.h b/lib/vhost/vhost.h
index d246538ca5..9521ae56da 100644
--- a/lib/vhost/vhost.h
+++ b/lib/vhost/vhost.h
@@ -354,7 +354,7 @@ struct vring_packed_desc_event {
 
 struct guest_page {
 	uint64_t guest_phys_addr;
-	uint64_t host_phys_addr;
+	uint64_t host_iova;
 	uint64_t host_user_addr;
 	uint64_t size;
 };
@@ -605,13 +605,13 @@ gpa_to_first_hpa(struct virtio_net *dev, uint64_t gpa,
 			if (gpa + gpa_size <=
 					page->guest_phys_addr + page->size) {
 				return gpa - page->guest_phys_addr +
-					page->host_phys_addr;
+					page->host_iova;
 			} else if (gpa < page->guest_phys_addr +
 						page->size) {
 				*hpa_size = page->guest_phys_addr +
 					page->size - gpa;
 				return gpa - page->guest_phys_addr +
-					page->host_phys_addr;
+					page->host_iova;
 			}
 		}
 	} else {
@@ -622,13 +622,13 @@ gpa_to_first_hpa(struct virtio_net *dev, uint64_t gpa,
 				if (gpa + gpa_size <=
 					page->guest_phys_addr + page->size) {
 					return gpa - page->guest_phys_addr +
-						page->host_phys_addr;
+						page->host_iova;
 				} else if (gpa < page->guest_phys_addr +
 							page->size) {
 					*hpa_size = page->guest_phys_addr +
 						page->size - gpa;
 					return gpa - page->guest_phys_addr +
-						page->host_phys_addr;
+						page->host_iova;
 				}
 			}
 		}
diff --git a/lib/vhost/vhost_user.c b/lib/vhost/vhost_user.c
index 6d888766b0..e2e56308b9 100644
--- a/lib/vhost/vhost_user.c
+++ b/lib/vhost/vhost_user.c
@@ -154,7 +154,7 @@ async_dma_map(struct virtio_net *dev, bool do_map)
 			page = &dev->guest_pages[i];
 			ret = rte_vfio_container_dma_map(RTE_VFIO_DEFAULT_CONTAINER_FD,
 							 page->host_user_addr,
-							 page->host_phys_addr,
+							 page->host_iova,
 							 page->size);
 			if (ret) {
 				/*
@@ -182,7 +182,7 @@ async_dma_map(struct virtio_net *dev, bool do_map)
 			page = &dev->guest_pages[i];
 			ret = rte_vfio_container_dma_unmap(RTE_VFIO_DEFAULT_CONTAINER_FD,
 							   page->host_user_addr,
-							   page->host_phys_addr,
+							   page->host_iova,
 							   page->size);
 			if (ret) {
 				/* like DMA map, ignore the kernel driver case when unmap. */
@@ -977,7 +977,7 @@ vhost_user_set_vring_base(struct virtio_net **pdev,
 
 static int
 add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
-		   uint64_t host_phys_addr, uint64_t host_user_addr, uint64_t size)
+		   uint64_t host_iova, uint64_t host_user_addr, uint64_t size)
 {
 	struct guest_page *page, *last_page;
 	struct guest_page *old_pages;
@@ -998,7 +998,7 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 	if (dev->nr_guest_pages > 0) {
 		last_page = &dev->guest_pages[dev->nr_guest_pages - 1];
 		/* merge if the two pages are continuous */
-		if (host_phys_addr == last_page->host_phys_addr + last_page->size
+		if (host_iova == last_page->host_iova + last_page->size
 		    && guest_phys_addr == last_page->guest_phys_addr + last_page->size
 		    && host_user_addr == last_page->host_user_addr + last_page->size) {
 			last_page->size += size;
@@ -1008,7 +1008,7 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 
 	page = &dev->guest_pages[dev->nr_guest_pages++];
 	page->guest_phys_addr = guest_phys_addr;
-	page->host_phys_addr  = host_phys_addr;
+	page->host_iova  = host_iova;
 	page->host_user_addr = host_user_addr;
 	page->size = size;
 
@@ -1022,14 +1022,14 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 	uint64_t reg_size = reg->size;
 	uint64_t host_user_addr  = reg->host_user_addr;
 	uint64_t guest_phys_addr = reg->guest_phys_addr;
-	uint64_t host_phys_addr;
+	uint64_t host_iova;
 	uint64_t size;
 
-	host_phys_addr = rte_mem_virt2iova((void *)(uintptr_t)host_user_addr);
+	host_iova = rte_mem_virt2iova((void *)(uintptr_t)host_user_addr);
 	size = page_size - (guest_phys_addr & (page_size - 1));
 	size = RTE_MIN(size, reg_size);
 
-	if (add_one_guest_page(dev, guest_phys_addr, host_phys_addr,
+	if (add_one_guest_page(dev, guest_phys_addr, host_iova,
 			       host_user_addr, size) < 0)
 		return -1;
 
@@ -1039,9 +1039,9 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 
 	while (reg_size > 0) {
 		size = RTE_MIN(reg_size, page_size);
-		host_phys_addr = rte_mem_virt2iova((void *)(uintptr_t)
+		host_iova = rte_mem_virt2iova((void *)(uintptr_t)
 						  host_user_addr);
-		if (add_one_guest_page(dev, guest_phys_addr, host_phys_addr,
+		if (add_one_guest_page(dev, guest_phys_addr, host_iova,
 				       host_user_addr, size) < 0)
 			return -1;
 
@@ -1073,11 +1073,11 @@ dump_guest_pages(struct virtio_net *dev)
 		VHOST_LOG_CONFIG(INFO,
 			"guest physical page region %u\n"
 			"\t guest_phys_addr: %" PRIx64 "\n"
-			"\t host_phys_addr : %" PRIx64 "\n"
+			"\t host_iova      : %" PRIx64 "\n"
 			"\t size           : %" PRIx64 "\n",
 			i,
 			page->guest_phys_addr,
-			page->host_phys_addr,
+			page->host_iova,
 			page->size);
 	}
 }
diff --git a/lib/vhost/virtio_net.c b/lib/vhost/virtio_net.c
index b3d954aab4..226dcc8b18 100644
--- a/lib/vhost/virtio_net.c
+++ b/lib/vhost/virtio_net.c
@@ -870,20 +870,21 @@ async_mbuf_to_desc_seg(struct virtio_net *dev, struct vhost_virtqueue *vq,
 	struct vhost_async *async = vq->async;
 	uint64_t mapped_len;
 	uint32_t buf_offset = 0;
-	void *hpa;
+	void *host_iova;
 
 	while (cpy_len) {
-		hpa = (void *)(uintptr_t)gpa_to_first_hpa(dev,
+		host_iova = (void *)(uintptr_t)gpa_to_first_hpa(dev,
 				buf_iova + buf_offset, cpy_len, &mapped_len);
-		if (unlikely(!hpa)) {
-			VHOST_LOG_DATA(ERR, "(%d) %s: failed to get hpa.\n", dev->vid, __func__);
+		if (unlikely(!host_iova)) {
+			VHOST_LOG_DATA(ERR, "(%d) %s: failed to get host iova.\n",
+				       dev->vid, __func__);
 			return -1;
 		}
 
 		if (unlikely(async_iter_add_iovec(async,
 						(void *)(uintptr_t)rte_pktmbuf_iova_offset(m,
 							mbuf_offset),
-						hpa, (size_t)mapped_len)))
+						host_iova, (size_t)mapped_len)))
 			return -1;
 
 		cpy_len -= (uint32_t)mapped_len;
-- 
2.17.1

[PATCH v4 0/2] vhost: fix async address mapping

[xuan.ding]

From: Xuan Ding <xuan.ding@intel.com>

This patchset fixes the issue of incorrect DMA mapping in PA mode.
Due to the ambiguity of host_phys_addr naming in the guest page
struct, rename it to host_iova.

v4:
* Fix the patch timezone.

v3:
* Fix some format issues.

v2:
* Change the order of patch.

Xuan Ding (2):
  vhost: fix field naming in guest page struct
  vhost: fix physical address mapping

 lib/vhost/vhost.h      |  11 ++--
 lib/vhost/vhost_user.c | 133 ++++++++++++++++++++---------------------
 lib/vhost/virtio_net.c |  11 ++--
 3 files changed, 76 insertions(+), 79 deletions(-)

-- 
2.17.1

[PATCH v4 1/2] vhost: fix field naming in guest page struct

[xuan.ding]

From: Xuan Ding <xuan.ding@intel.com>

This patch renames the host_phys_addr to host_iova in guest_page
struct. The host_phys_addr is iova, it depends on the DPDK
IOVA mode.

Fixes: e246896178e6 ("vhost: get guest/host physical address mappings")
Cc: stable@dpdk.org

Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
---
 lib/vhost/vhost.h      | 10 +++++-----
 lib/vhost/vhost_user.c | 20 ++++++++++----------
 lib/vhost/virtio_net.c | 11 ++++++-----
 3 files changed, 21 insertions(+), 20 deletions(-)

diff --git a/lib/vhost/vhost.h b/lib/vhost/vhost.h
index 1c2ee29600..ccac679c25 100644
--- a/lib/vhost/vhost.h
+++ b/lib/vhost/vhost.h
@@ -426,7 +426,7 @@ struct vring_packed_desc_event {
 
 struct guest_page {
 	uint64_t guest_phys_addr;
-	uint64_t host_phys_addr;
+	uint64_t host_iova;
 	uint64_t size;
 };
 
@@ -689,13 +689,13 @@ gpa_to_first_hpa(struct virtio_net *dev, uint64_t gpa,
 			if (gpa + gpa_size <=
 					page->guest_phys_addr + page->size) {
 				return gpa - page->guest_phys_addr +
-					page->host_phys_addr;
+					page->host_iova;
 			} else if (gpa < page->guest_phys_addr +
 						page->size) {
 				*hpa_size = page->guest_phys_addr +
 					page->size - gpa;
 				return gpa - page->guest_phys_addr +
-					page->host_phys_addr;
+					page->host_iova;
 			}
 		}
 	} else {
@@ -706,13 +706,13 @@ gpa_to_first_hpa(struct virtio_net *dev, uint64_t gpa,
 				if (gpa + gpa_size <=
 					page->guest_phys_addr + page->size) {
 					return gpa - page->guest_phys_addr +
-						page->host_phys_addr;
+						page->host_iova;
 				} else if (gpa < page->guest_phys_addr +
 							page->size) {
 					*hpa_size = page->guest_phys_addr +
 						page->size - gpa;
 					return gpa - page->guest_phys_addr +
-						page->host_phys_addr;
+						page->host_iova;
 				}
 			}
 		}
diff --git a/lib/vhost/vhost_user.c b/lib/vhost/vhost_user.c
index 1d3f89afd8..066adfb464 100644
--- a/lib/vhost/vhost_user.c
+++ b/lib/vhost/vhost_user.c
@@ -988,7 +988,7 @@ vhost_user_set_vring_base(struct virtio_net **pdev,
 
 static int
 add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
-		   uint64_t host_phys_addr, uint64_t size)
+		   uint64_t host_iova, uint64_t size)
 {
 	struct guest_page *page, *last_page;
 	struct guest_page *old_pages;
@@ -1009,7 +1009,7 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 	if (dev->nr_guest_pages > 0) {
 		last_page = &dev->guest_pages[dev->nr_guest_pages - 1];
 		/* merge if the two pages are continuous */
-		if (host_phys_addr == last_page->host_phys_addr +
+		if (host_iova == last_page->host_iova +
 				      last_page->size) {
 			last_page->size += size;
 			return 0;
@@ -1018,7 +1018,7 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 
 	page = &dev->guest_pages[dev->nr_guest_pages++];
 	page->guest_phys_addr = guest_phys_addr;
-	page->host_phys_addr  = host_phys_addr;
+	page->host_iova  = host_iova;
 	page->size = size;
 
 	return 0;
@@ -1031,14 +1031,14 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 	uint64_t reg_size = reg->size;
 	uint64_t host_user_addr  = reg->host_user_addr;
 	uint64_t guest_phys_addr = reg->guest_phys_addr;
-	uint64_t host_phys_addr;
+	uint64_t host_iova;
 	uint64_t size;
 
-	host_phys_addr = rte_mem_virt2iova((void *)(uintptr_t)host_user_addr);
+	host_iova = rte_mem_virt2iova((void *)(uintptr_t)host_user_addr);
 	size = page_size - (guest_phys_addr & (page_size - 1));
 	size = RTE_MIN(size, reg_size);
 
-	if (add_one_guest_page(dev, guest_phys_addr, host_phys_addr, size) < 0)
+	if (add_one_guest_page(dev, guest_phys_addr, host_iova, size) < 0)
 		return -1;
 
 	host_user_addr  += size;
@@ -1047,9 +1047,9 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 
 	while (reg_size > 0) {
 		size = RTE_MIN(reg_size, page_size);
-		host_phys_addr = rte_mem_virt2iova((void *)(uintptr_t)
+		host_iova = rte_mem_virt2iova((void *)(uintptr_t)
 						  host_user_addr);
-		if (add_one_guest_page(dev, guest_phys_addr, host_phys_addr,
+		if (add_one_guest_page(dev, guest_phys_addr, host_iova,
 				size) < 0)
 			return -1;
 
@@ -1082,8 +1082,8 @@ dump_guest_pages(struct virtio_net *dev)
 				dev->ifname, i);
 		VHOST_LOG_CONFIG(INFO, "(%s)\tguest_phys_addr: %" PRIx64 "\n",
 				dev->ifname, page->guest_phys_addr);
-		VHOST_LOG_CONFIG(INFO, "(%s)\thost_phys_addr : %" PRIx64 "\n",
-				dev->ifname, page->host_phys_addr);
+		VHOST_LOG_CONFIG(INFO, "(%s)\thost_iova : %" PRIx64 "\n",
+				dev->ifname, page->host_iova);
 		VHOST_LOG_CONFIG(INFO, "(%s)\tsize           : %" PRIx64 "\n",
 				dev->ifname, page->size);
 	}
diff --git a/lib/vhost/virtio_net.c b/lib/vhost/virtio_net.c
index 886e076b28..5f432b0d77 100644
--- a/lib/vhost/virtio_net.c
+++ b/lib/vhost/virtio_net.c
@@ -1004,20 +1004,21 @@ async_mbuf_to_desc_seg(struct virtio_net *dev, struct vhost_virtqueue *vq,
 	struct vhost_async *async = vq->async;
 	uint64_t mapped_len;
 	uint32_t buf_offset = 0;
-	void *hpa;
+	void *host_iova;
 
 	while (cpy_len) {
-		hpa = (void *)(uintptr_t)gpa_to_first_hpa(dev,
+		host_iova = (void *)(uintptr_t)gpa_to_first_hpa(dev,
 				buf_iova + buf_offset, cpy_len, &mapped_len);
-		if (unlikely(!hpa)) {
-			VHOST_LOG_DATA(ERR, "(%s) %s: failed to get hpa.\n", dev->ifname, __func__);
+		if (unlikely(!host_iova)) {
+			VHOST_LOG_DATA(ERR, "(%s) %s: failed to get host iova.\n",
+				       dev->ifname, __func__);
 			return -1;
 		}
 
 		if (unlikely(async_iter_add_iovec(dev, async,
 						(void *)(uintptr_t)rte_pktmbuf_iova_offset(m,
 							mbuf_offset),
-						hpa, (size_t)mapped_len)))
+						host_iova, (size_t)mapped_len)))
 			return -1;
 
 		cpy_len -= (uint32_t)mapped_len;
-- 
2.17.1

[PATCH v4 2/2] vhost: fix physical address mapping

[xuan.ding]

From: Xuan Ding <xuan.ding@intel.com>

When choosing IOVA as PA mode, IOVA is likely to be discontinuous,
which requires page by page mapping for DMA devices. To be consistent,
this patch implements page by page mapping instead of mapping at the
region granularity for both IOVA as VA and PA mode.

Fixes: 7c61fa08b716 ("vhost: enable IOMMU for async vhost")
Cc: stable@dpdk.org

Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Signed-off-by: Yuan Wang <yuanx.wang@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
---
 lib/vhost/vhost.h      |   1 +
 lib/vhost/vhost_user.c | 119 ++++++++++++++++++++---------------------
 2 files changed, 58 insertions(+), 62 deletions(-)

diff --git a/lib/vhost/vhost.h b/lib/vhost/vhost.h
index ccac679c25..21e1866a52 100644
--- a/lib/vhost/vhost.h
+++ b/lib/vhost/vhost.h
@@ -427,6 +427,7 @@ struct vring_packed_desc_event {
 struct guest_page {
 	uint64_t guest_phys_addr;
 	uint64_t host_iova;
+	uint64_t host_user_addr;
 	uint64_t size;
 };
 
diff --git a/lib/vhost/vhost_user.c b/lib/vhost/vhost_user.c
index 066adfb464..9d31d8840e 100644
--- a/lib/vhost/vhost_user.c
+++ b/lib/vhost/vhost_user.c
@@ -142,57 +142,57 @@ get_blk_size(int fd)
 	return ret == -1 ? (uint64_t)-1 : (uint64_t)stat.st_blksize;
 }
 
-static int
-async_dma_map(struct virtio_net *dev, struct rte_vhost_mem_region *region, bool do_map)
+static void
+async_dma_map(struct virtio_net *dev, bool do_map)
 {
-	uint64_t host_iova;
 	int ret = 0;
+	uint32_t i;
+	struct guest_page *page;
 
-	host_iova = rte_mem_virt2iova((void *)(uintptr_t)region->host_user_addr);
 	if (do_map) {
-		/* Add mapped region into the default container of DPDK. */
-		ret = rte_vfio_container_dma_map(RTE_VFIO_DEFAULT_CONTAINER_FD,
-						 region->host_user_addr,
-						 host_iova,
-						 region->size);
-		if (ret) {
-			/*
-			 * DMA device may bind with kernel driver, in this case,
-			 * we don't need to program IOMMU manually. However, if no
-			 * device is bound with vfio/uio in DPDK, and vfio kernel
-			 * module is loaded, the API will still be called and return
-			 * with ENODEV/ENOSUP.
-			 *
-			 * DPDK vfio only returns ENODEV/ENOSUP in very similar
-			 * situations(vfio either unsupported, or supported
-			 * but no devices found). Either way, no mappings could be
-			 * performed. We treat it as normal case in async path.
-			 */
-			if (rte_errno == ENODEV || rte_errno == ENOTSUP)
-				return 0;
-
-			VHOST_LOG_CONFIG(ERR, "(%s) DMA engine map failed\n", dev->ifname);
-			/* DMA mapping errors won't stop VHST_USER_SET_MEM_TABLE. */
-			return 0;
+		for (i = 0; i < dev->nr_guest_pages; i++) {
+			page = &dev->guest_pages[i];
+			ret = rte_vfio_container_dma_map(RTE_VFIO_DEFAULT_CONTAINER_FD,
+							 page->host_user_addr,
+							 page->host_iova,
+							 page->size);
+			if (ret) {
+				/*
+				 * DMA device may bind with kernel driver, in this case,
+				 * we don't need to program IOMMU manually. However, if no
+				 * device is bound with vfio/uio in DPDK, and vfio kernel
+				 * module is loaded, the API will still be called and return
+				 * with ENODEV.
+				 *
+				 * DPDK vfio only returns ENODEV in very similar situations
+				 * (vfio either unsupported, or supported but no devices found).
+				 * Either way, no mappings could be performed. We treat it as
+				 * normal case in async path. This is a workaround.
+				 */
+				if (rte_errno == ENODEV)
+					return;
+
+				/* DMA mapping errors won't stop VHOST_USER_SET_MEM_TABLE. */
+				VHOST_LOG_CONFIG(ERR, "DMA engine map failed\n");
+			}
 		}
 
 	} else {
-		/* Remove mapped region from the default container of DPDK. */
-		ret = rte_vfio_container_dma_unmap(RTE_VFIO_DEFAULT_CONTAINER_FD,
-						   region->host_user_addr,
-						   host_iova,
-						   region->size);
-		if (ret) {
-			/* like DMA map, ignore the kernel driver case when unmap. */
-			if (rte_errno == EINVAL)
-				return 0;
-
-			VHOST_LOG_CONFIG(ERR, "(%s) DMA engine unmap failed\n", dev->ifname);
-			return ret;
+		for (i = 0; i < dev->nr_guest_pages; i++) {
+			page = &dev->guest_pages[i];
+			ret = rte_vfio_container_dma_unmap(RTE_VFIO_DEFAULT_CONTAINER_FD,
+							   page->host_user_addr,
+							   page->host_iova,
+							   page->size);
+			if (ret) {
+				/* like DMA map, ignore the kernel driver case when unmap. */
+				if (rte_errno == EINVAL)
+					return;
+
+				VHOST_LOG_CONFIG(ERR, "DMA engine unmap failed\n");
+			}
 		}
 	}
-
-	return ret;
 }
 
 static void
@@ -204,12 +204,12 @@ free_mem_region(struct virtio_net *dev)
 	if (!dev || !dev->mem)
 		return;
 
+	if (dev->async_copy && rte_vfio_is_enabled("vfio"))
+		async_dma_map(dev, false);
+
 	for (i = 0; i < dev->mem->nregions; i++) {
 		reg = &dev->mem->regions[i];
 		if (reg->host_user_addr) {
-			if (dev->async_copy && rte_vfio_is_enabled("vfio"))
-				async_dma_map(dev, reg, false);
-
 			munmap(reg->mmap_addr, reg->mmap_size);
 			close(reg->fd);
 		}
@@ -988,7 +988,7 @@ vhost_user_set_vring_base(struct virtio_net **pdev,
 
 static int
 add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
-		   uint64_t host_iova, uint64_t size)
+		   uint64_t host_iova, uint64_t host_user_addr, uint64_t size)
 {
 	struct guest_page *page, *last_page;
 	struct guest_page *old_pages;
@@ -1000,7 +1000,7 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 					dev->max_guest_pages * sizeof(*page),
 					RTE_CACHE_LINE_SIZE);
 		if (dev->guest_pages == NULL) {
-			VHOST_LOG_CONFIG(ERR, "(%s) cannot realloc guest_pages\n", dev->ifname);
+			VHOST_LOG_CONFIG(ERR, "cannot realloc guest_pages\n");
 			rte_free(old_pages);
 			return -1;
 		}
@@ -1009,8 +1009,9 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 	if (dev->nr_guest_pages > 0) {
 		last_page = &dev->guest_pages[dev->nr_guest_pages - 1];
 		/* merge if the two pages are continuous */
-		if (host_iova == last_page->host_iova +
-				      last_page->size) {
+		if (host_iova == last_page->host_iova + last_page->size &&
+		    guest_phys_addr == last_page->guest_phys_addr + last_page->size &&
+		    host_user_addr == last_page->host_user_addr + last_page->size) {
 			last_page->size += size;
 			return 0;
 		}
@@ -1019,6 +1020,7 @@ add_one_guest_page(struct virtio_net *dev, uint64_t guest_phys_addr,
 	page = &dev->guest_pages[dev->nr_guest_pages++];
 	page->guest_phys_addr = guest_phys_addr;
 	page->host_iova  = host_iova;
+	page->host_user_addr = host_user_addr;
 	page->size = size;
 
 	return 0;
@@ -1038,7 +1040,8 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 	size = page_size - (guest_phys_addr & (page_size - 1));
 	size = RTE_MIN(size, reg_size);
 
-	if (add_one_guest_page(dev, guest_phys_addr, host_iova, size) < 0)
+	if (add_one_guest_page(dev, guest_phys_addr, host_iova,
+			       host_user_addr, size) < 0)
 		return -1;
 
 	host_user_addr  += size;
@@ -1050,7 +1053,7 @@ add_guest_pages(struct virtio_net *dev, struct rte_vhost_mem_region *reg,
 		host_iova = rte_mem_virt2iova((void *)(uintptr_t)
 						  host_user_addr);
 		if (add_one_guest_page(dev, guest_phys_addr, host_iova,
-				size) < 0)
+				       host_user_addr, size) < 0)
 			return -1;
 
 		host_user_addr  += size;
@@ -1226,7 +1229,6 @@ vhost_user_mmap_region(struct virtio_net *dev,
 	uint64_t mmap_size;
 	uint64_t alignment;
 	int populate;
-	int ret;
 
 	/* Check for memory_size + mmap_offset overflow */
 	if (mmap_offset >= -region->size) {
@@ -1283,16 +1285,6 @@ vhost_user_mmap_region(struct virtio_net *dev,
 					dev->ifname);
 			return -1;
 		}
-
-		if (rte_vfio_is_enabled("vfio")) {
-			ret = async_dma_map(dev, region, true);
-			if (ret) {
-				VHOST_LOG_CONFIG(ERR,
-					"(%s) configure IOMMU for DMA engine failed\n",
-					dev->ifname);
-				return -1;
-			}
-		}
 	}
 
 	VHOST_LOG_CONFIG(INFO, "(%s) guest memory region size: 0x%" PRIx64 "\n",
@@ -1429,6 +1421,9 @@ vhost_user_set_mem_table(struct virtio_net **pdev,
 		dev->mem->nregions++;
 	}
 
+	if (dev->async_copy && rte_vfio_is_enabled("vfio"))
+		async_dma_map(dev, true);
+
 	if (vhost_user_postcopy_register(dev, main_fd, ctx) < 0)
 		goto free_mem_table;
 
-- 
2.17.1

Re: [PATCH v4 0/2] vhost: fix async address mapping

[Maxime Coquelin]

On 2/16/22 03:28, xuan.ding@intel.com wrote:
> From: Xuan Ding <xuan.ding@intel.com>
> 
> This patchset fixes the issue of incorrect DMA mapping in PA mode.
> Due to the ambiguity of host_phys_addr naming in the guest page
> struct, rename it to host_iova.
> 
> v4:
> * Fix the patch timezone.
> 
> v3:
> * Fix some format issues.
> 
> v2:
> * Change the order of patch.
> 
> Xuan Ding (2):
>    vhost: fix field naming in guest page struct
>    vhost: fix physical address mapping
> 
>   lib/vhost/vhost.h      |  11 ++--
>   lib/vhost/vhost_user.c | 133 ++++++++++++++++++++---------------------
>   lib/vhost/virtio_net.c |  11 ++--
>   3 files changed, 76 insertions(+), 79 deletions(-)
> 

Applied to dpdk-next-virtio/main.

Thanks,
Maxime