[dpdk-dev] [PATCH 1/2] vhost: fix overflow on shadow used ring

[dpdk-dev] [PATCH 1/2] vhost: fix overflow on shadow used ring

[Tiwei Bie]

The shadow used ring's size is the same as the vq's size,
so we shouldn't try more than "vq size" times. Besides,
the element pointed by avail->idx isn't available to the
device, so we will return error when try "vq size" times.

Fixes: 24e4844048e1 ("vhost: unify Rx mergeable and non-mergeable paths")
Fixes: a922401f35cc ("vhost: add Rx support for packed ring")

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
---
 lib/librte_vhost/virtio_net.c | 37 ++++++++++++++++-------------------
 1 file changed, 17 insertions(+), 20 deletions(-)

diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index 5d4b97587..3b11b353c 100644
--- a/lib/librte_vhost/virtio_net.c
+++ b/lib/librte_vhost/virtio_net.c
@@ -415,13 +415,20 @@ reserve_avail_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
 	cur_idx  = vq->last_avail_idx;
 
 	if (rxvq_is_mergeable(dev))
-		max_tries = vq->size;
+		max_tries = vq->size - 1;
 	else
 		max_tries = 1;
 
 	while (size > 0) {
 		if (unlikely(cur_idx == avail_head))
 			return -1;
+		/*
+		 * if we tried all available ring items, and still
+		 * can't get enough buf, it means something abnormal
+		 * happened.
+		 */
+		if (unlikely(++tries > max_tries))
+			return -1;
 
 		if (unlikely(fill_vec_buf_split(dev, vq, cur_idx,
 						&vec_idx, buf_vec,
@@ -433,16 +440,7 @@ reserve_avail_buf_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
 		size -= len;
 
 		cur_idx++;
-		tries++;
 		*num_buffers += 1;
-
-		/*
-		 * if we tried all available ring items, and still
-		 * can't get enough buf, it means something abnormal
-		 * happened.
-		 */
-		if (unlikely(tries > max_tries))
-			return -1;
 	}
 
 	*nr_vec = vec_idx;
@@ -582,11 +580,19 @@ reserve_avail_buf_packed(struct virtio_net *dev, struct vhost_virtqueue *vq,
 	avail_idx = vq->last_avail_idx;
 
 	if (rxvq_is_mergeable(dev))
-		max_tries = vq->size;
+		max_tries = vq->size - 1;
 	else
 		max_tries = 1;
 
 	while (size > 0) {
+		/*
+		 * if we tried all available ring items, and still
+		 * can't get enough buf, it means something abnormal
+		 * happened.
+		 */
+		if (unlikely(++tries > max_tries))
+			return -1;
+
 		if (unlikely(fill_vec_buf_packed(dev, vq,
 						avail_idx, &desc_count,
 						buf_vec, &vec_idx,
@@ -603,16 +609,7 @@ reserve_avail_buf_packed(struct virtio_net *dev, struct vhost_virtqueue *vq,
 			avail_idx -= vq->size;
 
 		*nr_descs += desc_count;
-		tries++;
 		*num_buffers += 1;
-
-		/*
-		 * if we tried all available ring items, and still
-		 * can't get enough buf, it means something abnormal
-		 * happened.
-		 */
-		if (unlikely(tries > max_tries))
-			return -1;
 	}
 
 	*nr_vec = vec_idx;
-- 
2.18.0

[dpdk-dev] [PATCH 2/2] vhost: release the locks when failed to make RARP packet

[Tiwei Bie]

Fixes: eefac9536a90 ("vhost: postpone device creation until rings are mapped")
Cc: stable@dpdk.org

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
---
 lib/librte_vhost/virtio_net.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index 3b11b353c..b02651275 100644
--- a/lib/librte_vhost/virtio_net.c
+++ b/lib/librte_vhost/virtio_net.c
@@ -1626,7 +1626,8 @@ rte_vhost_dequeue_burst(int vid, uint16_t queue_id,
 		if (rarp_mbuf == NULL) {
 			RTE_LOG(ERR, VHOST_DATA,
 				"Failed to make RARP packet.\n");
-			return 0;
+			count = 0;
+			goto out;
 		}
 		count -= 1;
 	}
-- 
2.18.0

Re: [dpdk-dev] [PATCH 2/2] vhost: release the locks when failed to make RARP packet

[Jens Freimann]

On Thu, Jul 26, 2018 at 09:37:21AM +0800, Tiwei Bie wrote:
>Fixes: eefac9536a90 ("vhost: postpone device creation until rings are mapped")
>Cc: stable@dpdk.org
>
>Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
>---
> lib/librte_vhost/virtio_net.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>

Reviewed-by: Jens Freimann <jfreimann@redhat.com>

Re: [dpdk-dev] [PATCH 1/2] vhost: fix overflow on shadow used ring

[Jens Freimann]

On Thu, Jul 26, 2018 at 09:37:20AM +0800, Tiwei Bie wrote:
>The shadow used ring's size is the same as the vq's size,
>so we shouldn't try more than "vq size" times. Besides,
>the element pointed by avail->idx isn't available to the
>device, so we will return error when try "vq size" times.
>
>Fixes: 24e4844048e1 ("vhost: unify Rx mergeable and non-mergeable paths")
>Fixes: a922401f35cc ("vhost: add Rx support for packed ring")
>
>Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
>---
> lib/librte_vhost/virtio_net.c | 37 ++++++++++++++++-------------------
> 1 file changed, 17 insertions(+), 20 deletions(-)
>

Reviewed-by: Jens Freimann <jfreiman@redhat.com>

Re: [dpdk-dev] [PATCH 1/2] vhost: fix overflow on shadow used ring

[Tiwei Bie]

On Thu, Jul 26, 2018 at 09:37:20AM +0800, Tiwei Bie wrote:
> The shadow used ring's size is the same as the vq's size,
> so we shouldn't try more than "vq size" times. Besides,
> the element pointed by avail->idx isn't available to the
> device, so we will return error when try "vq size" times.
> 
> Fixes: 24e4844048e1 ("vhost: unify Rx mergeable and non-mergeable paths")
> Fixes: a922401f35cc ("vhost: add Rx support for packed ring")
> 
> Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>

Series applied to dpdk-next-virtio/master.