On Wed, Mar 21, 2018 at 09:45:59AM +0100, Jens Freimann wrote:

On Mon, Mar 19, 2018 at 07:02:56PM +0800, Tiwei Bie wrote:

quoted

On Fri, Mar 16, 2018 at 04:21:16PM +0100, Jens Freimann wrote:
[...]

quoted

+static inline uint32_t __attribute__((always_inline))
+vhost_enqueue_burst_packed(struct virtio_net *dev, uint16_t queue_id,
+	      struct rte_mbuf **pkts, uint32_t count)
+{
+	struct vhost_virtqueue *vq;
+	struct vring_desc_packed *descs;
+	uint16_t idx;
+	uint16_t mask;
+	uint16_t i;
+
+	vq = dev->virtqueue[queue_id];
+
+	rte_spinlock_lock(&vq->access_lock);
+
+	if (unlikely(vq->enabled == 0)) {
+		i = 0;
+		goto out_access_unlock;
+	}
+
+	if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM))
+		vhost_user_iotlb_rd_lock(vq);
+
+	descs = vq->desc_packed;
+	mask = vq->size - 1;

Queue size may not be a power of 2 in packed ring.

yes, you are right. my patches don't support this yet, but I plan to
add it in the next version.

Yeah, thanks. Otherwise, because this is vhost code, and
the vq->size is set by virtio driver. So when above code
works with a virtio driver which uses a ring size that's
not a power of 2, such assumption will cause problems.

Thanks

Thanks for the review!

regards,
Jens