On 10/11/22 05:08, Cheng Jiang wrote:
When the packet receiving failure and the DMA ring full occur simultaneously in the asynchronous vhost, the slot_idx needs to be decreased by 1. For packed virtqueue, the slot index should be ring_size - 1, if the slot_idx is currently 0, since the ring size is not necessarily the power of 2. Fixes: 84d5204310d7 ("vhost: support async dequeue for split ring") Fixes: fe8477ebbd94 ("vhost: support async packed ring dequeue") Cc: sta...@dpdk.org Signed-off-by: Cheng Jiang <cheng1.ji...@intel.com> --- lib/vhost/virtio_net.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/lib/vhost/virtio_net.c b/lib/vhost/virtio_net.c index 457ac2e92a..efebd063d7 100644 --- a/lib/vhost/virtio_net.c +++ b/lib/vhost/virtio_net.c @@ -3457,6 +3457,7 @@ virtio_dev_tx_async_split(struct virtio_net *dev, struct vhost_virtqueue *vq, allocerr_warned = true; } dropped = true; + slot_idx--; break; }@@ -3647,6 +3648,12 @@ virtio_dev_tx_async_packed(struct virtio_net *dev, struct vhost_virtqueue *vq,if (unlikely(virtio_dev_tx_async_single_packed(dev, vq, mbuf_pool, pkt, slot_idx, legacy_ol_flags))) { rte_pktmbuf_free_bulk(&pkts_prealloc[pkt_idx], count - pkt_idx); + + if (slot_idx == 0) + slot_idx = vq->size - 1; + else + slot_idx--; + break; }@@ -3674,8 +3681,13 @@ virtio_dev_tx_async_packed(struct virtio_net *dev, struct vhost_virtqueue *vq,async->buffer_idx_packed += vq->size - pkt_err;while (pkt_err-- > 0) {- rte_pktmbuf_free(pkts_info[slot_idx % vq->size].mbuf); - slot_idx--; + rte_pktmbuf_free(pkts_info[slot_idx].mbuf); + descs_err += pkts_info[slot_idx].descs; + + if (slot_idx == 0) + slot_idx = vq->size - 1; + else + slot_idx--; }/* recover available ring */
Reviewed-by: Maxime Coquelin <maxime.coque...@redhat.com> Thanks, Maxime
