[dpdk-dev] [PATCH v6 07/15] vhost: extract split ring handling from Rx and Tx functions
Tiwei Bie
tiwei.bie at intel.com
Wed Jul 4 08:51:36 CEST 2018
On Mon, Jul 02, 2018 at 10:16:21AM +0200, Maxime Coquelin wrote:
[...]
> +
> + if (unlikely(vq->access_ok == 0))
> + if (unlikely(vring_translate(dev, vq) < 0))
> + goto out;
> +
> +
Please just keep one blank line.
> + /*
> + * Construct a RARP broadcast packet, and inject it to the "pkts"
> + * array, to looks like that guest actually send such packet.
> + *
> + * Check user_send_rarp() for more information.
> + *
> + * broadcast_rarp shares a cacheline in the virtio_net structure
> + * with some fields that are accessed during enqueue and
> + * rte_atomic16_cmpset() causes a write if using cmpxchg. This could
> + * result in false sharing between enqueue and dequeue.
> + *
> + * Prevent unnecessary false sharing by reading broadcast_rarp first
> + * and only performing cmpset if the read indicates it is likely to
> + * be set.
> + */
> +
Please remove above blank line.
> + if (unlikely(rte_atomic16_read(&dev->broadcast_rarp) &&
> + rte_atomic16_cmpset((volatile uint16_t *)
> + &dev->broadcast_rarp.cnt, 1, 0))) {
> +
> + rarp_mbuf = rte_net_make_rarp_packet(mbuf_pool, &dev->mac);
> + if (rarp_mbuf == NULL) {
> + RTE_LOG(ERR, VHOST_DATA,
> + "Failed to make RARP packet.\n");
> + return 0;
> + }
> + count -= 1;
> + }
> +
> + count = virtio_dev_tx_split(dev, vq, mbuf_pool, pkts, count);
> +
> out:
> if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM))
> vhost_user_iotlb_rd_unlock(vq);
> @@ -1200,10 +1222,10 @@ rte_vhost_dequeue_burst(int vid, uint16_t queue_id,
> * Inject it to the head of "pkts" array, so that switch's mac
> * learning table will get updated first.
> */
> - memmove(&pkts[1], pkts, i * sizeof(struct rte_mbuf *));
> + memmove(&pkts[1], pkts, count * sizeof(struct rte_mbuf *));
> pkts[0] = rarp_mbuf;
> - i += 1;
> + count += 1;
> }
>
> - return i;
> + return count;
> }
> --
> 2.14.4
>
More information about the dev
mailing list