[v3,2/4] net/virtio: fix segmented packet issue in mergeable Rx path
Checks
Commit Message
After having dequeued a burst of descriptors, there may be a
need to dequeue a few more if the last packet was segmented
and not complete. When it happens, the extra segments were
not properly attached to the mbuf chain, and so were lost.
Also, head segment data_len field is wrongly summed with
the length of all the segments of the chain.
This patch fixes both the mbuf chaining and head segment's
data_len field
Fixes: bcac5aa207f8 ("net/virtio: improve batching in mergeable path")
Cc: stable@dpdk.org
Reported-by: Yaroslav Brustinov <ybrustin@cisco.com>
Reviewed-by: David Marchand <david.marchand@redhat.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
---
drivers/net/virtio/virtio_rxtx.c | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
@@ -1613,7 +1613,7 @@ virtio_recv_mergeable_pkts(void *rx_queue,
struct virtqueue *vq = rxvq->vq;
struct virtio_hw *hw = vq->hw;
struct rte_mbuf *rxm;
- struct rte_mbuf *prev;
+ struct rte_mbuf *prev = NULL;
uint16_t nb_used, num, nb_rx = 0;
uint32_t len[VIRTIO_MBUF_BURST_SZ];
struct rte_mbuf *rcv_pkts[VIRTIO_MBUF_BURST_SZ];
@@ -1700,7 +1700,6 @@ virtio_recv_mergeable_pkts(void *rx_queue,
rxm->data_len = (uint16_t)(len[i]);
rx_pkts[nb_rx]->pkt_len += (uint32_t)(len[i]);
- rx_pkts[nb_rx]->data_len += (uint16_t)(len[i]);
if (prev)
prev->next = rxm;
@@ -1720,7 +1719,6 @@ virtio_recv_mergeable_pkts(void *rx_queue,
uint16_t rcv_cnt = RTE_MIN((uint16_t)seg_res,
VIRTIO_MBUF_BURST_SZ);
- prev = rcv_pkts[nb_rx];
if (likely(VIRTQUEUE_NUSED(vq) >= rcv_cnt)) {
virtio_rmb(hw->weak_barriers);
num = virtqueue_dequeue_burst_rx(vq, rcv_pkts, len,
@@ -1737,7 +1735,6 @@ virtio_recv_mergeable_pkts(void *rx_queue,
prev->next = rxm;
prev = rxm;
rx_pkts[nb_rx]->pkt_len += len[extra_idx];
- rx_pkts[nb_rx]->data_len += len[extra_idx];
extra_idx += 1;
};
seg_res -= rcv_cnt;