[ovs-dev] [PATCH 4/5] dpif-netdev: Assign PMD for failed pinned rxqs.

[David Marchand]

On Fri, Jun 4, 2021 at 11:19 PM Kevin Traynor <ktraynor at redhat.com> wrote:
>
> Previously, if pmd-rxq-affinity was used to pin an rxq to
> a core that was not in pmd-cpu-mask the rxq was not polled
> and the user received a warning.
>
> Now that pinned and non-pinned rxqs are assigned to PMDs in
> a common call to rxq scheduling, if an invalid core is
> selected in pmd-rxq-affinity the rxq can be assigned an
> available PMD (if any).
>
> A warning will still be logged as the requested core could
> not be used.
>
> Signed-off-by: Kevin Traynor <ktraynor at redhat.com>
> ---
>  Documentation/topics/dpdk/pmd.rst |  6 +++---
>  lib/dpif-netdev.c                 | 30 ++++++++++++++++++++++++++++--
>  tests/pmd.at                      |  5 ++++-
>  3 files changed, 35 insertions(+), 6 deletions(-)
>
> diff --git a/Documentation/topics/dpdk/pmd.rst b/Documentation/topics/dpdk/pmd.rst
> index d1c45cdfb..29ba53954 100644
> --- a/Documentation/topics/dpdk/pmd.rst
> +++ b/Documentation/topics/dpdk/pmd.rst
> @@ -108,7 +108,7 @@ means that this thread will only poll the *pinned* Rx queues.
>
>     If there are no *non-isolated* PMD threads, *non-pinned* RX queues will not
> -   be polled. Also, if the provided ``<core-id>`` is not available (e.g. the
> -   ``<core-id>`` is not in ``pmd-cpu-mask``), the RX queue will not be polled
> -   by any PMD thread.
> +   be polled. If the provided ``<core-id>`` is not available (e.g. the
> +   ``<core-id>`` is not in ``pmd-cpu-mask``), the RX queue will be assigned to
> +   a *non-isolated* PMD, that will remain *non-isolated*.
>
>  If ``pmd-rxq-affinity`` is not set for Rx queues, they will be assigned to PMDs
> diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
> index 61e0a516f..377573233 100644
> --- a/lib/dpif-netdev.c
> +++ b/lib/dpif-netdev.c
> @@ -5027,4 +5027,25 @@ find_sched_pmd_by_pmd(struct sched_numa_list *numa_list,
>  }
>
> +static struct sched_pmd *
> +find_sched_pmd_by_rxq(struct sched_numa_list *numa_list,
> +                      struct dp_netdev_rxq *rxq)
> +{
> +    struct sched_numa *numa;
> +
> +    HMAP_FOR_EACH (numa, node, &numa_list->numas) {
> +        for (unsigned i = 0; i < numa->n_pmds; i++) {
> +            struct sched_pmd *sched_pmd;
> +
> +            sched_pmd = &numa->pmds[i];
> +            for (int k = 0; k < sched_pmd->n_rxq; k++) {
> +                if (sched_pmd->rxqs[k] == rxq) {
> +                    return sched_pmd;
> +                }
> +            }
> +        }
> +    }
> +    return NULL;
> +}
> +
>  static struct sched_numa *
>  sched_numa_list_find_numa(struct sched_numa_list *numa_list,
> @@ -5408,5 +5429,6 @@ sched_numa_list_schedule(struct sched_numa_list *numa_list,
>                              "Core %2u cannot be pinned with "
>                              "port \'%s\' rx queue %d. Use pmd-cpu-mask to "
> -                            "enable a pmd on core %u.",
> +                            "enable a pmd on core %u. An alternative core "
> +                            "will be assigned.",
>                              rxq->core_id,
>                              netdev_rxq_get_name(rxq->rx),

To continue with my suggestion on patch 1 on only pushing unpinned
queues to the rxqs array, you could add this rxq in rxqs[] ...

> @@ -5453,5 +5475,9 @@ sched_numa_list_schedule(struct sched_numa_list *numa_list,
>
>          if (rxq->core_id != OVS_CORE_UNSPEC) {
> -            continue;
> +             /* This rxq should have been pinned, check it was. */
> +            sched_pmd = find_sched_pmd_by_rxq(numa_list, rxq);
> +            if (sched_pmd && sched_pmd->pmd->core_id == rxq->core_id) {
> +                continue;
> +            }

... and this check is unnneded.

WDYT?

>          }
>
> diff --git a/tests/pmd.at b/tests/pmd.at
> index 78105bf45..55977632a 100644
> --- a/tests/pmd.at
> +++ b/tests/pmd.at
> @@ -552,7 +552,10 @@ AT_CHECK([ovs-vsctl set Open_vSwitch . other_config:pmd-cpu-mask=6])
>
>  dnl We removed the cores requested by some queues from pmd-cpu-mask.
> -dnl Those queues will not be polled.
> +dnl Those queues will be polled by remaining non-isolated pmds.
>  AT_CHECK([ovs-appctl dpif-netdev/pmd-rxq-show | parse_pmd_rxq_show], [0], [dnl
> +p1 0 0 1
> +p1 1 0 1
>  p1 2 0 2
> +p1 3 0 1
>  ])
>
> --
> 2.31.1
>



-- 
David Marchand