On Fri, May 29, 2015 at 12:42:00PM +0100, Mark D. Gray wrote:
> From: Daniele Di Proietto <diproiet...@vmware.com>
>
> Non pmd threads have a core_id == UINT32_MAX, while queue ids used by
> netdevs range from 0 to the number of CPUs. Therefore core ids cannot
> be used directly to select a queue.
>
> This commit introduces a simple mapping to fix the problem: non pmd
> threads use queue 0, pmd threads on core 0 to N use queues 1 to N+1
>
> Fixes: d5c199ea7ff7 ("netdev-dpdk: Properly support non pmd threads.")
>
> Reported-by: 차은호 <eunho....@atto-research.com
> Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
> Signed-off-by: Mark D. Gray <mark.d.g...@intel.com>
> ---
> lib/dpif-netdev.c | 20 ++++++++++++++++----
> lib/netdev-dpdk.c | 10 ++++++----
> 2 files changed, 22 insertions(+), 8 deletions(-)
>
> diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
> index 76d1003..de700f7 100644
> --- a/lib/dpif-netdev.c
> +++ b/lib/dpif-netdev.c
> @@ -1068,7 +1068,7 @@ do_add_port(struct dp_netdev *dp, const char *devname,
> const char *type,
> }
> /* There can only be ovs_numa_get_n_cores() pmd threads,
> * so creates a txq for each. */
> - error = netdev_set_multiq(netdev, n_cores, dp->n_dpdk_rxqs);
> + error = netdev_set_multiq(netdev, n_cores + 1, dp->n_dpdk_rxqs);
> if (error && (error != EOPNOTSUPP)) {
> VLOG_ERR("%s, cannot set multiq", devname);
> return errno;
> @@ -2402,7 +2402,8 @@ dpif_netdev_pmd_set(struct dpif *dpif, unsigned int
> n_rxqs, const char *cmask)
> }
>
> /* Sets the new rx queue config. */
> - err = netdev_set_multiq(port->netdev, ovs_numa_get_n_cores(),
> + err = netdev_set_multiq(port->netdev,
> + ovs_numa_get_n_cores() + 1,
> n_rxqs);
> if (err && (err != EOPNOTSUPP)) {
> VLOG_ERR("Failed to set dpdk interface %s rx_queue to:"
> @@ -3328,8 +3329,18 @@ dpif_netdev_register_upcall_cb(struct dpif *dpif,
> upcall_callback *cb,
> dp->upcall_cb = cb;
> }
>
> +static int
> +core_id_to_qid(unsigned core_id)
> +{
> + if (core_id != NON_PMD_CORE_ID) {
> + return core_id + 1;
> + } else {
> + return 0;
> + }
> +}
> +
> static void
> -dp_netdev_drop_packets(struct dp_packet ** packets, int cnt, bool may_steal)
> +dp_netdev_drop_packets(struct dp_packet **packets, int cnt, bool may_steal)
> {
> if (may_steal) {
> int i;
> @@ -3387,7 +3398,8 @@ dp_execute_cb(void *aux_, struct dp_packet **packets,
> int cnt,
> case OVS_ACTION_ATTR_OUTPUT:
> p = dp_netdev_lookup_port(dp, u32_to_odp(nl_attr_get_u32(a)));
> if (OVS_LIKELY(p)) {
> - netdev_send(p->netdev, pmd->core_id, packets, cnt, may_steal);
> + netdev_send(p->netdev, core_id_to_qid(pmd->core_id), packets,
> cnt,
> + may_steal);
> return;
> }
> break;
> diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
> index 63243d8..1d20d98 100644
> --- a/lib/netdev-dpdk.c
> +++ b/lib/netdev-dpdk.c
> @@ -892,9 +892,11 @@ netdev_dpdk_rxq_recv(struct netdev_rxq *rxq_, struct
> dp_packet **packets,
> int nb_rx;
>
> /* There is only one tx queue for this core. Do not flush other
> - * queueus. */
> + * queues. */
> if (rxq_->queue_id == rte_lcore_id()) {
> - dpdk_queue_flush(dev, rxq_->queue_id);
> + /* TX queue 0 is the nonpmd queue. Each core owns TX queue
> + * number = core_id + 1 */
> + dpdk_queue_flush(dev, rxq_->queue_id + 1);
I found this patch a bit hard to follow without the commit log, so
maybe move all the '+ 1' to functions close to each other and add
a comment explaining how the mapping works? The core_id_to_qid()
is a good example.
> }
>
> nb_rx = rte_eth_rx_burst(rx->port_id, rxq_->queue_id,
> @@ -1070,8 +1072,8 @@ dpdk_do_tx_copy(struct netdev *netdev, int qid, struct
> dp_packet **pkts,
> if (dev->type == DPDK_DEV_VHOST) {
> __netdev_dpdk_vhost_send(netdev, (struct dp_packet **) mbufs,
> newcnt, true);
> } else {
> - dpdk_queue_pkts(dev, qid, mbufs, newcnt);
> - dpdk_queue_flush(dev, qid);
> + dpdk_queue_pkts(dev, qid, mbufs, newcnt);
> + dpdk_queue_flush(dev, qid);
This doesn't look right.
fbl
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
