thr3ads.net - Virtualization - [PATCH 0/5] virtio_ring: per virtqueue DMA device [Jan 2023]

If this information is useful, please help other people find it:
Share via:

Jason Wang

2023-Jan-11 06:28 UTC

[PATCH 0/5] virtio_ring: per virtqueue DMA device

Hi All:

In some cases, the virtqueue could be backed by different devices. One
example is that in the case of vDPA some parent may emualte virtqueue
via vringh. In this case, it would be wrong if we stick with the
physical DMA device for software emulated device, since there's no
easy way for vringh to know about the hardware IOMMU mappings.

So this series tries to introduce per virtqueue DMA device, then
software virtqueues can utilize the transport specific method to
assign appropirate DMA device.

This fixes the crash of mlx5_vdpa + virtio_vdpa when platform IOMMU is
enabled but not in the passthrough mode. The reason for the crash is
that the virito_ring tries to map the control virtqueue into platform
IOMMU but the vringh assumes a direct mapping (PA as IOVA). This is
fixed by advetise the vDPA device that doesnt do DMA (without a DMA
ops). So DMA API can go with the direct mapping then the vringh will
be happy since mlx5_vdpa assuems a direct/identical mapping by
default.

Please review.

Thanks

Jason Wang (5):
  virtio_ring: per virtqueue dma device
  vdpa: introduce get_vq_dma_device()
  virtio-vdpa: support per vq dma device
  vdpa: set dma mask for vDPA device
  vdpa: mlx5: support per virtqueue dma device

 drivers/vdpa/mlx5/net/mlx5_vnet.c |  11 +++
 drivers/vdpa/vdpa.c               |   5 ++
 drivers/virtio/virtio_ring.c      | 133 +++++++++++++++++++++---------
 drivers/virtio/virtio_vdpa.c      |  13 ++-
 include/linux/vdpa.h              |   6 ++
 include/linux/virtio_ring.h       |  16 ++++
 6 files changed, 141 insertions(+), 43 deletions(-)

-- 
2.25.1

Jason Wang

2023-Jan-11 06:28 UTC

head link

[PATCH 1/5] virtio_ring: per virtqueue dma device

This patch introduces a per virtqueue dma device. This will be used
for virtio devices whose virtqueue are backed by different underlayer
devices.

One example is the vDPA that where the control virtqueue could be
implemented through software mediation.

Some of the work are actually done before since the helper like
vring_dma_device(). This work left are:

- Let vring_dma_device() return the per virtqueue dma device instead
  of the vdev's parent.
- Allow passing a dma_device when creating the virtqueue through a new
  helper, old vring creation helper will keep using vdev's parent.

Signed-off-by: Jason Wang <jasowang at redhat.com>
---
 drivers/virtio/virtio_ring.c | 133 ++++++++++++++++++++++++-----------
 include/linux/virtio_ring.h  |  16 +++++
 2 files changed, 109 insertions(+), 40 deletions(-)

diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
index 723c4e29e1d3..41144b5246a8 100644
--- a/drivers/virtio/virtio_ring.c
+++ b/drivers/virtio/virtio_ring.c
@@ -202,6 +202,9 @@ struct vring_virtqueue {
 	/* DMA, allocation, and size information */
 	bool we_own_ring;
 
+	/* Device used for doing DMA */
+	struct device *dma_dev;
+
 #ifdef DEBUG
 	/* They're supposed to lock for us. */
 	unsigned int in_use;
@@ -219,7 +222,8 @@ static struct virtqueue *__vring_new_virtqueue(unsigned int
index,
 					       bool context,
 					       bool (*notify)(struct virtqueue *),
 					       void (*callback)(struct virtqueue *),
-					       const char *name);
+					       const char *name,
+					       struct device *dma_dev);
 static struct vring_desc_extra *vring_alloc_desc_extra(unsigned int num);
 static void vring_free(struct virtqueue *_vq);
 
@@ -297,10 +301,11 @@ size_t virtio_max_dma_size(struct virtio_device *vdev)
 EXPORT_SYMBOL_GPL(virtio_max_dma_size);
 
 static void *vring_alloc_queue(struct virtio_device *vdev, size_t size,
-			      dma_addr_t *dma_handle, gfp_t flag)
+			       dma_addr_t *dma_handle, gfp_t flag,
+			       struct device *dma_dev)
 {
 	if (vring_use_dma_api(vdev)) {
-		return dma_alloc_coherent(vdev->dev.parent, size,
+		return dma_alloc_coherent(dma_dev, size,
 					  dma_handle, flag);
 	} else {
 		void *queue = alloc_pages_exact(PAGE_ALIGN(size), flag);
@@ -330,10 +335,11 @@ static void *vring_alloc_queue(struct virtio_device *vdev,
size_t size,
 }
 
 static void vring_free_queue(struct virtio_device *vdev, size_t size,
-			     void *queue, dma_addr_t dma_handle)
+			     void *queue, dma_addr_t dma_handle,
+			     struct device *dma_dev)
 {
 	if (vring_use_dma_api(vdev))
-		dma_free_coherent(vdev->dev.parent, size, queue, dma_handle);
+		dma_free_coherent(dma_dev, size, queue, dma_handle);
 	else
 		free_pages_exact(queue, PAGE_ALIGN(size));
 }
@@ -341,11 +347,11 @@ static void vring_free_queue(struct virtio_device *vdev,
size_t size,
 /*
  * The DMA ops on various arches are rather gnarly right now, and
  * making all of the arch DMA ops work on the vring device itself
- * is a mess.  For now, we use the parent device for DMA ops.
+ * is a mess.
  */
 static inline struct device *vring_dma_dev(const struct vring_virtqueue *vq)
 {
-	return vq->vq.vdev->dev.parent;
+	return vq->dma_dev;
 }
 
 /* Map one sg entry. */
@@ -1032,11 +1038,12 @@ static int vring_alloc_state_extra_split(struct
vring_virtqueue_split *vring_spl
 }
 
 static void vring_free_split(struct vring_virtqueue_split *vring_split,
-			     struct virtio_device *vdev)
+			     struct virtio_device *vdev, struct device *dma_dev)
 {
 	vring_free_queue(vdev, vring_split->queue_size_in_bytes,
 			 vring_split->vring.desc,
-			 vring_split->queue_dma_addr);
+			 vring_split->queue_dma_addr,
+			 dma_dev);
 
 	kfree(vring_split->desc_state);
 	kfree(vring_split->desc_extra);
@@ -1046,7 +1053,8 @@ static int vring_alloc_queue_split(struct
vring_virtqueue_split *vring_split,
 				   struct virtio_device *vdev,
 				   u32 num,
 				   unsigned int vring_align,
-				   bool may_reduce_num)
+				   bool may_reduce_num,
+				   struct device *dma_dev)
 {
 	void *queue = NULL;
 	dma_addr_t dma_addr;
@@ -1061,7 +1069,8 @@ static int vring_alloc_queue_split(struct
vring_virtqueue_split *vring_split,
 	for (; num && vring_size(num, vring_align) > PAGE_SIZE; num /= 2) {
 		queue = vring_alloc_queue(vdev, vring_size(num, vring_align),
 					  &dma_addr,
-					  GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO);
+					  GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO,
+					  dma_dev);
 		if (queue)
 			break;
 		if (!may_reduce_num)
@@ -1074,7 +1083,8 @@ static int vring_alloc_queue_split(struct
vring_virtqueue_split *vring_split,
 	if (!queue) {
 		/* Try to get a single page. You are my only hope! */
 		queue = vring_alloc_queue(vdev, vring_size(num, vring_align),
-					  &dma_addr, GFP_KERNEL | __GFP_ZERO);
+					  &dma_addr, GFP_KERNEL | __GFP_ZERO,
+					  dma_dev);
 	}
 	if (!queue)
 		return -ENOMEM;
@@ -1100,21 +1110,22 @@ static struct virtqueue *vring_create_virtqueue_split(
 	bool context,
 	bool (*notify)(struct virtqueue *),
 	void (*callback)(struct virtqueue *),
-	const char *name)
+	const char *name,
+	struct device *dma_dev)
 {
 	struct vring_virtqueue_split vring_split = {};
 	struct virtqueue *vq;
 	int err;
 
 	err = vring_alloc_queue_split(&vring_split, vdev, num, vring_align,
-				      may_reduce_num);
+				      may_reduce_num, dma_dev);
 	if (err)
 		return NULL;
 
 	vq = __vring_new_virtqueue(index, &vring_split, vdev, weak_barriers,
-				   context, notify, callback, name);
+				   context, notify, callback, name, dma_dev);
 	if (!vq) {
-		vring_free_split(&vring_split, vdev);
+		vring_free_split(&vring_split, vdev, dma_dev);
 		return NULL;
 	}
 
@@ -1132,7 +1143,8 @@ static int virtqueue_resize_split(struct virtqueue *_vq,
u32 num)
 
 	err = vring_alloc_queue_split(&vring_split, vdev, num,
 				      vq->split.vring_align,
-				      vq->split.may_reduce_num);
+				      vq->split.may_reduce_num,
+				      vring_dma_dev(vq));
 	if (err)
 		goto err;
 
@@ -1150,7 +1162,7 @@ static int virtqueue_resize_split(struct virtqueue *_vq,
u32 num)
 	return 0;
 
 err_state_extra:
-	vring_free_split(&vring_split, vdev);
+	vring_free_split(&vring_split, vdev, vring_dma_dev(vq));
 err:
 	virtqueue_reinit_split(vq);
 	return -ENOMEM;
@@ -1841,22 +1853,26 @@ static struct vring_desc_extra
*vring_alloc_desc_extra(unsigned int num)
 }
 
 static void vring_free_packed(struct vring_virtqueue_packed *vring_packed,
-			      struct virtio_device *vdev)
+			      struct virtio_device *vdev,
+			      struct device *dma_dev)
 {
 	if (vring_packed->vring.desc)
 		vring_free_queue(vdev, vring_packed->ring_size_in_bytes,
 				 vring_packed->vring.desc,
-				 vring_packed->ring_dma_addr);
+				 vring_packed->ring_dma_addr,
+				 dma_dev);
 
 	if (vring_packed->vring.driver)
 		vring_free_queue(vdev, vring_packed->event_size_in_bytes,
 				 vring_packed->vring.driver,
-				 vring_packed->driver_event_dma_addr);
+				 vring_packed->driver_event_dma_addr,
+				 dma_dev);
 
 	if (vring_packed->vring.device)
 		vring_free_queue(vdev, vring_packed->event_size_in_bytes,
 				 vring_packed->vring.device,
-				 vring_packed->device_event_dma_addr);
+				 vring_packed->device_event_dma_addr,
+				 dma_dev);
 
 	kfree(vring_packed->desc_state);
 	kfree(vring_packed->desc_extra);
@@ -1864,7 +1880,7 @@ static void vring_free_packed(struct
vring_virtqueue_packed *vring_packed,
 
 static int vring_alloc_queue_packed(struct vring_virtqueue_packed
*vring_packed,
 				    struct virtio_device *vdev,
-				    u32 num)
+				    u32 num, struct device *dma_dev)
 {
 	struct vring_packed_desc *ring;
 	struct vring_packed_desc_event *driver, *device;
@@ -1875,7 +1891,8 @@ static int vring_alloc_queue_packed(struct
vring_virtqueue_packed *vring_packed,
 
 	ring = vring_alloc_queue(vdev, ring_size_in_bytes,
 				 &ring_dma_addr,
-				 GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO);
+				 GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO,
+				 dma_dev);
 	if (!ring)
 		goto err;
 
@@ -1887,7 +1904,8 @@ static int vring_alloc_queue_packed(struct
vring_virtqueue_packed *vring_packed,
 
 	driver = vring_alloc_queue(vdev, event_size_in_bytes,
 				   &driver_event_dma_addr,
-				   GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO);
+				   GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO,
+				   dma_dev);
 	if (!driver)
 		goto err;
 
@@ -1897,7 +1915,8 @@ static int vring_alloc_queue_packed(struct
vring_virtqueue_packed *vring_packed,
 
 	device = vring_alloc_queue(vdev, event_size_in_bytes,
 				   &device_event_dma_addr,
-				   GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO);
+				   GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO,
+				   dma_dev);
 	if (!device)
 		goto err;
 
@@ -1909,7 +1928,7 @@ static int vring_alloc_queue_packed(struct
vring_virtqueue_packed *vring_packed,
 	return 0;
 
 err:
-	vring_free_packed(vring_packed, vdev);
+	vring_free_packed(vring_packed, vdev, dma_dev);
 	return -ENOMEM;
 }
 
@@ -1987,13 +2006,14 @@ static struct virtqueue *vring_create_virtqueue_packed(
 	bool context,
 	bool (*notify)(struct virtqueue *),
 	void (*callback)(struct virtqueue *),
-	const char *name)
+	const char *name,
+	struct device *dma_dev)
 {
 	struct vring_virtqueue_packed vring_packed = {};
 	struct vring_virtqueue *vq;
 	int err;
 
-	if (vring_alloc_queue_packed(&vring_packed, vdev, num))
+	if (vring_alloc_queue_packed(&vring_packed, vdev, num, dma_dev))
 		goto err_ring;
 
 	vq = kmalloc(sizeof(*vq), GFP_KERNEL);
@@ -2014,6 +2034,7 @@ static struct virtqueue *vring_create_virtqueue_packed(
 	vq->broken = false;
 #endif
 	vq->packed_ring = true;
+	vq->dma_dev = dma_dev;
 	vq->use_dma_api = vring_use_dma_api(vdev);
 
 	vq->indirect = virtio_has_feature(vdev, VIRTIO_RING_F_INDIRECT_DESC)
&&
@@ -2040,7 +2061,7 @@ static struct virtqueue *vring_create_virtqueue_packed(
 err_state_extra:
 	kfree(vq);
 err_vq:
-	vring_free_packed(&vring_packed, vdev);
+	vring_free_packed(&vring_packed, vdev, dma_dev);
 err_ring:
 	return NULL;
 }
@@ -2052,7 +2073,7 @@ static int virtqueue_resize_packed(struct virtqueue *_vq,
u32 num)
 	struct virtio_device *vdev = _vq->vdev;
 	int err;
 
-	if (vring_alloc_queue_packed(&vring_packed, vdev, num))
+	if (vring_alloc_queue_packed(&vring_packed, vdev, num, vring_dma_dev(vq)))
 		goto err_ring;
 
 	err = vring_alloc_state_extra_packed(&vring_packed);
@@ -2069,7 +2090,7 @@ static int virtqueue_resize_packed(struct virtqueue *_vq,
u32 num)
 	return 0;
 
 err_state_extra:
-	vring_free_packed(&vring_packed, vdev);
+	vring_free_packed(&vring_packed, vdev, vring_dma_dev(vq));
 err_ring:
 	virtqueue_reinit_packed(vq);
 	return -ENOMEM;
@@ -2481,7 +2502,8 @@ static struct virtqueue *__vring_new_virtqueue(unsigned
int index,
 					       bool context,
 					       bool (*notify)(struct virtqueue *),
 					       void (*callback)(struct virtqueue *),
-					       const char *name)
+					       const char *name,
+					       struct device *dma_dev)
 {
 	struct vring_virtqueue *vq;
 	int err;
@@ -2507,6 +2529,7 @@ static struct virtqueue *__vring_new_virtqueue(unsigned
int index,
 #else
 	vq->broken = false;
 #endif
+	vq->dma_dev = dma_dev;
 	vq->use_dma_api = vring_use_dma_api(vdev);
 
 	vq->indirect = virtio_has_feature(vdev, VIRTIO_RING_F_INDIRECT_DESC)
&&
@@ -2549,14 +2572,39 @@ struct virtqueue *vring_create_virtqueue(
 	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED))
 		return vring_create_virtqueue_packed(index, num, vring_align,
 				vdev, weak_barriers, may_reduce_num,
-				context, notify, callback, name);
+				context, notify, callback, name, vdev->dev.parent);
 
 	return vring_create_virtqueue_split(index, num, vring_align,
 			vdev, weak_barriers, may_reduce_num,
-			context, notify, callback, name);
+			context, notify, callback, name, vdev->dev.parent);
 }
 EXPORT_SYMBOL_GPL(vring_create_virtqueue);
 
+struct virtqueue *vring_create_virtqueue_dma(
+	unsigned int index,
+	unsigned int num,
+	unsigned int vring_align,
+	struct virtio_device *vdev,
+	bool weak_barriers,
+	bool may_reduce_num,
+	bool context,
+	bool (*notify)(struct virtqueue *),
+	void (*callback)(struct virtqueue *),
+	const char *name,
+	struct device *dma_dev)
+{
+
+	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED))
+		return vring_create_virtqueue_packed(index, num, vring_align,
+				vdev, weak_barriers, may_reduce_num,
+				context, notify, callback, name, dma_dev);
+
+	return vring_create_virtqueue_split(index, num, vring_align,
+			vdev, weak_barriers, may_reduce_num,
+			context, notify, callback, name, dma_dev);
+}
+EXPORT_SYMBOL_GPL(vring_create_virtqueue_dma);
+
 /**
  * virtqueue_resize - resize the vring of vq
  * @_vq: the struct virtqueue we're talking about.
@@ -2645,7 +2693,8 @@ struct virtqueue *vring_new_virtqueue(unsigned int index,
 
 	vring_init(&vring_split.vring, num, pages, vring_align);
 	return __vring_new_virtqueue(index, &vring_split, vdev, weak_barriers,
-				     context, notify, callback, name);
+				     context, notify, callback, name,
+				     vdev->dev.parent);
 }
 EXPORT_SYMBOL_GPL(vring_new_virtqueue);
 
@@ -2658,17 +2707,20 @@ static void vring_free(struct virtqueue *_vq)
 			vring_free_queue(vq->vq.vdev,
 					 vq->packed.ring_size_in_bytes,
 					 vq->packed.vring.desc,
-					 vq->packed.ring_dma_addr);
+					 vq->packed.ring_dma_addr,
+					 vring_dma_dev(vq));
 
 			vring_free_queue(vq->vq.vdev,
 					 vq->packed.event_size_in_bytes,
 					 vq->packed.vring.driver,
-					 vq->packed.driver_event_dma_addr);
+					 vq->packed.driver_event_dma_addr,
+					 vring_dma_dev(vq));
 
 			vring_free_queue(vq->vq.vdev,
 					 vq->packed.event_size_in_bytes,
 					 vq->packed.vring.device,
-					 vq->packed.device_event_dma_addr);
+					 vq->packed.device_event_dma_addr,
+					 vring_dma_dev(vq));
 
 			kfree(vq->packed.desc_state);
 			kfree(vq->packed.desc_extra);
@@ -2676,7 +2728,8 @@ static void vring_free(struct virtqueue *_vq)
 			vring_free_queue(vq->vq.vdev,
 					 vq->split.queue_size_in_bytes,
 					 vq->split.vring.desc,
-					 vq->split.queue_dma_addr);
+					 vq->split.queue_dma_addr,
+					 vring_dma_dev(vq));
 		}
 	}
 	if (!vq->packed_ring) {
diff --git a/include/linux/virtio_ring.h b/include/linux/virtio_ring.h
index 8b8af1a38991..8b95b69ef694 100644
--- a/include/linux/virtio_ring.h
+++ b/include/linux/virtio_ring.h
@@ -76,6 +76,22 @@ struct virtqueue *vring_create_virtqueue(unsigned int index,
 					 void (*callback)(struct virtqueue *vq),
 					 const char *name);
 
+/*
+ * Creates a virtqueue and allocates the descriptor ring with per
+ * virtqueue DMA device.
+ */
+struct virtqueue *vring_create_virtqueue_dma(unsigned int index,
+					     unsigned int num,
+					     unsigned int vring_align,
+					     struct virtio_device *vdev,
+					     bool weak_barriers,
+					     bool may_reduce_num,
+					     bool ctx,
+					     bool (*notify)(struct virtqueue *vq),
+					     void (*callback)(struct virtqueue *vq),
+					     const char *name,
+					     struct device *dma_dev);
+
 /*
  * Creates a virtqueue with a standard layout but a caller-allocated
  * ring.
-- 
2.25.1

Jason Wang

2023-Jan-11 06:28 UTC

head link

[PATCH 2/5] vdpa: introduce get_vq_dma_device()

This patch introduces a new method to query the dma device that is use
for a specific virtqueue.

Signed-off-by: Jason Wang <jasowang at redhat.com>
---
 include/linux/vdpa.h | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/include/linux/vdpa.h b/include/linux/vdpa.h
index 6d0f5e4e82c2..3ec13aee35f5 100644
--- a/include/linux/vdpa.h
+++ b/include/linux/vdpa.h
@@ -282,6 +282,11 @@ struct vdpa_map_file {
  *				@iova: iova to be unmapped
  *				@size: size of the area
  *				Returns integer: success (0) or error (< 0)
+ * @get_vq_dma_dev:		Get the dma device for a specific
+ *				virtqueue (optional)
+ *				@vdev: vdpa device
+ *				@idx: virtqueue index
+ *				Returns pointer to structure device or error (NULL)
  * @free:			Free resources that belongs to vDPA (optional)
  *				@vdev: vdpa device
  */
@@ -341,6 +346,7 @@ struct vdpa_config_ops {
 			 u64 iova, u64 size);
 	int (*set_group_asid)(struct vdpa_device *vdev, unsigned int group,
 			      unsigned int asid);
+	struct device *(*get_vq_dma_dev)(struct vdpa_device *vdev, u16 idx);
 
 	/* Free device resources */
 	void (*free)(struct vdpa_device *vdev);
-- 
2.25.1

Jason Wang

2023-Jan-11 06:28 UTC

head link

[PATCH 3/5] virtio-vdpa: support per vq dma device

This patch adds the support of per vq dma device for virito-vDPA. vDPA
parents then are allowed to use different DMA devices. This is useful
for the parents that have software or emulated virtqueues.

Signed-off-by: Jason Wang <jasowang at redhat.com>
---
 drivers/virtio/virtio_vdpa.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
index 9670cc79371d..d7f5af62ddaa 100644
--- a/drivers/virtio/virtio_vdpa.c
+++ b/drivers/virtio/virtio_vdpa.c
@@ -135,6 +135,7 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned
int index,
 {
 	struct virtio_vdpa_device *vd_dev = to_virtio_vdpa_device(vdev);
 	struct vdpa_device *vdpa = vd_get_vdpa(vdev);
+	struct device *dma_dev;
 	const struct vdpa_config_ops *ops = vdpa->config;
 	struct virtio_vdpa_vq_info *info;
 	struct vdpa_callback cb;
@@ -175,9 +176,15 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned
int index,
 
 	/* Create the vring */
 	align = ops->get_vq_align(vdpa);
-	vq = vring_create_virtqueue(index, max_num, align, vdev,
-				    true, may_reduce_num, ctx,
-				    virtio_vdpa_notify, callback, name);
+
+	if (ops->get_vq_dma_dev)
+		dma_dev = ops->get_vq_dma_dev(vdpa, index);
+	else
+		dma_dev = vdpa_get_dma_dev(vdpa);
+	vq = vring_create_virtqueue_dma(index, max_num, align, vdev,
+					true, may_reduce_num, ctx,
+					virtio_vdpa_notify, callback,
+					name, dma_dev);
 	if (!vq) {
 		err = -ENOMEM;
 		goto error_new_virtqueue;
-- 
2.25.1

Jason Wang

2023-Jan-11 06:28 UTC

head link

[PATCH 4/5] vdpa: set dma mask for vDPA device

Setting DMA mask for vDPA device in case that there are virtqueue that
is not backed by DMA so the vDPA device could be advertised as the DMA
device that is used by DMA API for software emulated virtqueues.

Signed-off-by: Jason Wang <jasowang at redhat.com>
---
 drivers/vdpa/vdpa.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
index 8ef7aa1365cc..6821b2850bbb 100644
--- a/drivers/vdpa/vdpa.c
+++ b/drivers/vdpa/vdpa.c
@@ -39,6 +39,11 @@ static int vdpa_dev_probe(struct device *d)
 	u32 max_num, min_num = 1;
 	int ret = 0;
 
+	d->dma_mask = &d->coherent_dma_mask;
+	ret = dma_set_mask_and_coherent(d, DMA_BIT_MASK(64));
+	if (ret)
+		return ret;
+
 	max_num = ops->get_vq_num_max(vdev);
 	if (ops->get_vq_num_min)
 		min_num = ops->get_vq_num_min(vdev);
-- 
2.25.1

Jason Wang

2023-Jan-11 06:28 UTC

head link

[PATCH 5/5] vdpa: mlx5: support per virtqueue dma device

This patch implements per virtqueue dma device for mlx5_vdpa. This is
needed for virtio_vdpa to work for CVQ which is backed by vringh but
not DMA. We simply advertise the vDPA device itself as the DMA device
for CVQ then DMA API can simply use PA so the identical mapping for
CVQ can still be used. Otherwise the identical (1:1) mapping won't
work when platform IOMMU is enabled since the IOVA is allocated on
demand which is not necessarily the PA.

This fixes the following crash when mlx5 vDPA device is bound to
virtio-vdpa with platform IOMMU enabled but not in passthrough mode:

BUG: unable to handle page fault for address: ff2fb3063deb1002
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 1393001067 P4D 1393002067 PUD 0
Oops: 0000 [#1] PREEMPT SMP NOPTI
CPU: 55 PID: 8923 Comm: kworker/u112:3 Kdump: loaded Not tainted 6.1.0+ #7
Hardware name: Dell Inc. PowerEdge R750/0PJ80M, BIOS 1.5.4 12/17/2021
Workqueue: mlx5_vdpa_wq mlx5_cvq_kick_handler [mlx5_vdpa]
RIP: 0010:vringh_getdesc_iotlb+0x93/0x1d0 [vringh]
Code: 14 25 40 ef 01 00 83 82 c0 0a 00 00 01 48 2b 05 93 5a 1b ea 8b 4c 24 14 48
c1 f8 06 48 c1 e0 0c 48 03 05 90 5a 1b ea 48 01 c8 <0f> b7 00 83 aa c0 0a
00 00 01 65 ff 0d bc e4 41 3f 0f 84 05 01 00
RSP: 0018:ff46821ba664fdf8 EFLAGS: 00010282
RAX: ff2fb3063deb1002 RBX: 0000000000000a20 RCX: 0000000000000002
RDX: ff2fb318d2f94380 RSI: 0000000000000002 RDI: 0000000000000001
RBP: ff2fb3065e832410 R08: ff46821ba664fe00 R09: 0000000000000001
R10: 0000000000000000 R11: 000000000000000d R12: ff2fb3065e832488
R13: ff2fb3065e8324a8 R14: ff2fb3065e8324c8 R15: ff2fb3065e8324a8
FS:  0000000000000000(0000) GS:ff2fb3257fac0000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: ff2fb3063deb1002 CR3: 0000001392010006 CR4: 0000000000771ee0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
PKRU: 55555554
Call Trace:
<TASK>
  mlx5_cvq_kick_handler+0x89/0x2b0 [mlx5_vdpa]
  process_one_work+0x1e2/0x3b0
  ? rescuer_thread+0x390/0x390
  worker_thread+0x50/0x3a0
  ? rescuer_thread+0x390/0x390
  kthread+0xd6/0x100
  ? kthread_complete_and_exit+0x20/0x20
  ret_from_fork+0x1f/0x30
  </TASK>

Signed-off-by: Jason Wang <jasowang at redhat.com>
---
 drivers/vdpa/mlx5/net/mlx5_vnet.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c
b/drivers/vdpa/mlx5/net/mlx5_vnet.c
index 6632651b1e54..b0e74c25bf48 100644
--- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
+++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
@@ -2682,6 +2682,16 @@ static int mlx5_vdpa_set_map(struct vdpa_device *vdev,
unsigned int asid,
 	return err;
 }
 
+struct device *mlx5_get_vq_dma_dev(struct vdpa_device *vdev, u16 idx)
+{
+	struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev);
+
+	if (is_ctrl_vq_idx(mvdev, idx))
+		return &vdev->dev;
+
+	return mvdev->vdev.dma_dev;
+}
+
 static void mlx5_vdpa_free(struct vdpa_device *vdev)
 {
 	struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev);
@@ -2897,6 +2907,7 @@ static const struct vdpa_config_ops mlx5_vdpa_ops = {
 	.get_generation = mlx5_vdpa_get_generation,
 	.set_map = mlx5_vdpa_set_map,
 	.set_group_asid = mlx5_set_group_asid,
+	.get_vq_dma_dev = mlx5_get_vq_dma_dev,
 	.free = mlx5_vdpa_free,
 	.suspend = mlx5_vdpa_suspend,
 };
-- 
2.25.1

Virtualization - Jan 2023 - [PATCH 0/5] virtio_ring: per virtqueue DMA device

[PATCH 0/5] virtio_ring: per virtqueue DMA device

[PATCH 1/5] virtio_ring: per virtqueue dma device

[PATCH 2/5] vdpa: introduce get_vq_dma_device()

[PATCH 3/5] virtio-vdpa: support per vq dma device

[PATCH 4/5] vdpa: set dma mask for vDPA device

[PATCH 5/5] vdpa: mlx5: support per virtqueue dma device