[ovs-dev] [PATCH v3 2/9] doc: Add "PMD" topic document

Kevin Traynor ktraynor at redhat.com
Thu Apr 19 15:08:03 UTC 2018


On 04/19/2018 01:57 PM, Stephen Finucane wrote:
> This continues the breakup of the huge DPDK "howto" into smaller
> components. There are a couple of related changes included, such as
> using "Rx queue" instead of "rxq" and noting how Tx queues cannot be
> configured.
> 
> Signed-off-by: Stephen Finucane <stephen at that.guru>
> ---
> v3:
> - Add new files to automake.mk
> - Add additional 'versionchanged' information for 'pmd-rxq-show' command
> v2:
> - Add cross-references from 'pmd' doc to 'vhost-user' and 'phy' docs
> - Add 'versionchanged' warning about automatic assignment of Rx queues
> - Add a 'todo' to describe Tx queue behavior
> ---
>  Documentation/automake.mk                |   1 +
>  Documentation/howto/dpdk.rst             |  86 -----------------
>  Documentation/topics/dpdk/index.rst      |   1 +
>  Documentation/topics/dpdk/phy.rst        |  12 +++
>  Documentation/topics/dpdk/pmd.rst        | 161 +++++++++++++++++++++++++++++++
>  Documentation/topics/dpdk/vhost-user.rst |  17 ++--
>  6 files changed, 183 insertions(+), 95 deletions(-)
>  create mode 100644 Documentation/topics/dpdk/pmd.rst
> 
> diff --git a/Documentation/automake.mk b/Documentation/automake.mk
> index cd8f3dc0d..878de4349 100644
> --- a/Documentation/automake.mk
> +++ b/Documentation/automake.mk
> @@ -35,6 +35,7 @@ DOC_SOURCE = \
>  	Documentation/topics/design.rst \
>  	Documentation/topics/dpdk/index.rst \
>  	Documentation/topics/dpdk/phy.rst \
> +	Documentation/topics/dpdk/pmd.rst \
>  	Documentation/topics/dpdk/ring.rst \
>  	Documentation/topics/dpdk/vhost-user.rst \
>  	Documentation/topics/testing.rst \
> diff --git a/Documentation/howto/dpdk.rst b/Documentation/howto/dpdk.rst
> index 79b626c76..388728363 100644
> --- a/Documentation/howto/dpdk.rst
> +++ b/Documentation/howto/dpdk.rst
> @@ -81,92 +81,6 @@ To stop ovs-vswitchd & delete bridge, run::
>      $ ovs-appctl -t ovsdb-server exit
>      $ ovs-vsctl del-br br0
>  
> -PMD Thread Statistics
> ----------------------
> -
> -To show current stats::
> -
> -    $ ovs-appctl dpif-netdev/pmd-stats-show
> -
> -To clear previous stats::
> -
> -    $ ovs-appctl dpif-netdev/pmd-stats-clear
> -
> -Port/RXQ Assigment to PMD Threads
> ----------------------------------
> -
> -To show port/rxq assignment::
> -
> -    $ ovs-appctl dpif-netdev/pmd-rxq-show
> -
> -To change default rxq assignment to pmd threads, rxqs may be manually pinned to
> -desired cores using::
> -
> -    $ ovs-vsctl set Interface <iface> \
> -        other_config:pmd-rxq-affinity=<rxq-affinity-list>
> -
> -where:
> -
> -- ``<rxq-affinity-list>`` is a CSV list of ``<queue-id>:<core-id>`` values
> -
> -For example::
> -
> -    $ ovs-vsctl set interface dpdk-p0 options:n_rxq=4 \
> -        other_config:pmd-rxq-affinity="0:3,1:7,3:8"
> -
> -This will ensure:
> -
> -- Queue #0 pinned to core 3
> -- Queue #1 pinned to core 7
> -- Queue #2 not pinned
> -- Queue #3 pinned to core 8
> -
> -After that PMD threads on cores where RX queues was pinned will become
> -``isolated``. This means that this thread will poll only pinned RX queues.
> -
> -.. warning::
> -  If there are no ``non-isolated`` PMD threads, ``non-pinned`` RX queues will
> -  not be polled. Also, if provided ``core_id`` is not available (ex. this
> -  ``core_id`` not in ``pmd-cpu-mask``), RX queue will not be polled by any PMD
> -  thread.
> -
> -If pmd-rxq-affinity is not set for rxqs, they will be assigned to pmds (cores)
> -automatically. The processing cycles that have been stored for each rxq
> -will be used where known to assign rxqs to pmd based on a round robin of the
> -sorted rxqs.
> -
> -For example, in the case where here there are 5 rxqs and 3 cores (e.g. 3,7,8)
> -available, and the measured usage of core cycles per rxq over the last
> -interval is seen to be:
> -
> -- Queue #0: 30%
> -- Queue #1: 80%
> -- Queue #3: 60%
> -- Queue #4: 70%
> -- Queue #5: 10%
> -
> -The rxqs will be assigned to cores 3,7,8 in the following order:
> -
> -Core 3: Q1 (80%) |
> -Core 7: Q4 (70%) | Q5 (10%)
> -core 8: Q3 (60%) | Q0 (30%)
> -
> -To see the current measured usage history of pmd core cycles for each rxq::
> -
> -    $ ovs-appctl dpif-netdev/pmd-rxq-show
> -
> -.. note::
> -
> -  A history of one minute is recorded and shown for each rxq to allow for
> -  traffic pattern spikes. An rxq's pmd core cycles usage changes due to traffic
> -  pattern or reconfig changes will take one minute before they are fully
> -  reflected in the stats.
> -
> -Rxq to pmds assignment takes place whenever there are configuration changes
> -or can be triggered by using::
> -
> -    $ ovs-appctl dpif-netdev/pmd-rxq-rebalance
> -
>  QoS
>  ---
>  
> diff --git a/Documentation/topics/dpdk/index.rst b/Documentation/topics/dpdk/index.rst
> index 5f836a6e9..dfde88377 100644
> --- a/Documentation/topics/dpdk/index.rst
> +++ b/Documentation/topics/dpdk/index.rst
> @@ -31,3 +31,4 @@ The DPDK Datapath
>     phy
>     vhost-user
>     ring
> +   pmd
> diff --git a/Documentation/topics/dpdk/phy.rst b/Documentation/topics/dpdk/phy.rst
> index a3f8b475c..ad191dad0 100644
> --- a/Documentation/topics/dpdk/phy.rst
> +++ b/Documentation/topics/dpdk/phy.rst
> @@ -113,3 +113,15 @@ tool::
>  For more information, refer to the `DPDK documentation <dpdk-drivers>`__.
>  
>  .. _dpdk-drivers: http://dpdk.org/doc/guides/linux_gsg/linux_drivers.html
> +
> +.. _dpdk-phy-multiqueue:
> +
> +Multiqueue
> +----------
> +
> +Poll Mode Driver (PMD) threads are the threads that do the heavy lifting for
> +the DPDK datapath. Correct configuration of PMD threads and the Rx queues they
> +utilize is a requirement in order to deliver the high-performance possible with
> +DPDK acceleration. It is possible to configure multiple Rx queues for ``dpdk``
> +ports, thus ensuring this is not a bottleneck for performance. For information
> +on configuring PMD threads, refer to :doc:`pmd`.
> diff --git a/Documentation/topics/dpdk/pmd.rst b/Documentation/topics/dpdk/pmd.rst
> new file mode 100644
> index 000000000..a652720e2
> --- /dev/null
> +++ b/Documentation/topics/dpdk/pmd.rst
> @@ -0,0 +1,161 @@
> +..
> +      Licensed under the Apache License, Version 2.0 (the "License"); you may
> +      not use this file except in compliance with the License. You may obtain
> +      a copy of the License at
> +
> +          http://www.apache.org/licenses/LICENSE-2.0
> +
> +      Unless required by applicable law or agreed to in writing, software
> +      distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
> +      WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
> +      License for the specific language governing permissions and limitations
> +      under the License.
> +
> +      Convention for heading levels in Open vSwitch documentation:
> +
> +      =======  Heading 0 (reserved for the title in a document)
> +      -------  Heading 1
> +      ~~~~~~~  Heading 2
> +      +++++++  Heading 3
> +      '''''''  Heading 4
> +
> +      Avoid deeper levels because they do not render well.
> +
> +===========
> +PMD Threads
> +===========
> +
> +Poll Mode Driver (PMD) threads are the threads that do the heavy lifting for
> +the DPDK datapath and perform tasks such as continuous polling of input ports
> +for packets, classifying packets once received, and executing actions on the
> +packets once they are classified.
> +
> +PMD threads utilize Receive (Rx) and Transmit (Tx) queues, commonly known as
> +*rxq*\s and *txq*\s. While Tx queue configuration happens automatically, Rx
> +queues can be configured by the user. This can happen in one of two ways:
> +
> +- For physical interfaces, configuration is done using the
> +  :program:`ovs-appctl` utility.
> +
> +- For virtual interfaces, configuration is done using the :program:`ovs-appctl`
> +  utility, but this configuration must be reflected in the guest configuration
> +  (e.g. QEMU command line arguments).
> +
> +The :program:`ovs-appctl` utility also provides a number of commands for
> +querying PMD threads and their respective queues. This, and all of the above,
> +is discussed here.
> +
> +.. todo::
> +
> +   Add an overview of Tx queues including numbers created, how they relate to
> +   PMD threads, etc.
> +
> +PMD Thread Statistics
> +---------------------
> +
> +To show current stats::
> +
> +    $ ovs-appctl dpif-netdev/pmd-stats-show
> +
> +To clear previous stats::
> +
> +    $ ovs-appctl dpif-netdev/pmd-stats-clear
> +
> +Port/Rx Queue Assigment to PMD Threads
> +--------------------------------------
> +
> +.. todo::
> +
> +   This needs a more detailed overview of *why* this should be done, along with
> +   the impact on things like NUMA affinity.
> +
> +Correct configuration of PMD threads and the Rx queues they utilize is a
> +requirement in order to achieve maximum performance. This is particularly true
> +for enabling things like multiqueue for :ref:`physical <dpdk-phy-multiqueue>`
> +and :ref:`vhost-user <dpdk-vhost-user>` interfaces.
> +
> +To show port/Rx queue assignment::
> +
> +    $ ovs-appctl dpif-netdev/pmd-rxq-show
> +
> +Rx queues may be manually pinned to cores. This will change the default Rx
> +queue assignment to PMD threads::
> +
> +    $ ovs-vsctl set Interface <iface> \
> +        other_config:pmd-rxq-affinity=<rxq-affinity-list>
> +
> +where:
> +
> +- ``<rxq-affinity-list>`` is a CSV list of ``<queue-id>:<core-id>`` values
> +
> +For example::
> +
> +    $ ovs-vsctl set interface dpdk-p0 options:n_rxq=4 \
> +        other_config:pmd-rxq-affinity="0:3,1:7,3:8"
> +
> +This will ensure there are *4* Rx queues and that these queues are configured
> +like so:
> +
> +- Queue #0 pinned to core 3
> +- Queue #1 pinned to core 7
> +- Queue #2 not pinned
> +- Queue #3 pinned to core 8
> +
> +PMD threads on cores where Rx queues are *pinned* will become *isolated*. This
> +means that this thread will only poll the *pinned* Rx queues.
> +
> +.. warning::
> +
> +   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.
> +
> +If ``pmd-rxq-affinity`` is not set for Rx queues, they will be assigned to PMDs
> +(cores) automatically. Where known, the processing cycles that have been stored
> +for each Rx queue will be used to assign Rx queue to PMDs based on a round
> +robin of the sorted Rx queues. For example, take the following example, where
> +there are five Rx queues and three cores - 3, 7, and 8 - available and the
> +measured usage of core cycles per Rx queue over the last interval is seen to
> +be:
> +
> +- Queue #0: 30%
> +- Queue #1: 80%
> +- Queue #3: 60%
> +- Queue #4: 70%
> +- Queue #5: 10%
> +
> +The Rx queues will be assigned to the cores in the following order::
> +
> +    Core 3: Q1 (80%) |
> +    Core 7: Q4 (70%) | Q5 (10%)
> +    Core 8: Q3 (60%) | Q0 (30%)
> +
> +To see the current measured usage history of PMD core cycles for each Rx
> +queue::
> +
> +    $ ovs-appctl dpif-netdev/pmd-rxq-show
> +
> +.. note::
> +
> +   A history of one minute is recorded and shown for each Rx queue to allow for
> +   traffic pattern spikes. Any changes in the Rx queue's PMD core cycles usage,
> +   due to traffic pattern or reconfig changes, will take one minute to be fully
> +   reflected in the stats.
> +
> +Rx queue to PMD assignment takes place whenever there are configuration changes
> +or can be triggered by using::
> +
> +    $ ovs-appctl dpif-netdev/pmd-rxq-rebalance
> +
> +.. versionchanged:: 2.8.0
> +
> +   Automatic assignment of Rx queues to PMDs and the two related commands,
> +   ``pmd-rxq-show`` and ``pmd-rxq-rebalance``, were added in OVS 2.8.0. Prior

pmd-rxq-show added in OVS 2.6, modified in 2.9 as noted below
pmd-rxq-rebalance added in OVS 2.9

> +   to this, behavior was round-robin and processing cycles were not taken into
> +   consideration. Tracking for stats was not available.
> +
> +.. versionchanged:: 2.9.0
> +
> +   The output of ``pmd-rxq-show`` was modified to include utilization as a
> +   percentage.
> diff --git a/Documentation/topics/dpdk/vhost-user.rst b/Documentation/topics/dpdk/vhost-user.rst
> index ca8a3289f..6f794f296 100644
> --- a/Documentation/topics/dpdk/vhost-user.rst
> +++ b/Documentation/topics/dpdk/vhost-user.rst
> @@ -130,11 +130,10 @@ an additional set of parameters::
>      -netdev type=vhost-user,id=mynet2,chardev=char2,vhostforce
>      -device virtio-net-pci,mac=00:00:00:00:00:02,netdev=mynet2
>  
> -In addition,       QEMU must allocate the VM's memory on hugetlbfs. vhost-user
> -ports access a virtio-net device's virtual rings and packet buffers mapping the
> -VM's physical memory on hugetlbfs. To enable vhost-user ports to map the VM's
> -memory into their process address space, pass the following parameters to
> -QEMU::
> +In addition, QEMU must allocate the VM's memory on hugetlbfs. vhost-user ports
> +access a virtio-net device's virtual rings and packet buffers mapping the VM's
> +physical memory on hugetlbfs. To enable vhost-user ports to map the VM's memory
> +into their process address space, pass the following parameters to QEMU::
>  
>      -object memory-backend-file,id=mem,size=4096M,mem-path=/dev/hugepages,share=on
>      -numa node,memdev=mem -mem-prealloc
> @@ -154,18 +153,18 @@ where:
>    The number of vectors, which is ``$q`` * 2 + 2
>  
>  The vhost-user interface will be automatically reconfigured with required
> -number of rx and tx queues after connection of virtio device.  Manual
> +number of Rx and Tx queues after connection of virtio device.  Manual
>  configuration of ``n_rxq`` is not supported because OVS will work properly only
>  if ``n_rxq`` will match number of queues configured in QEMU.
>  
> -A least 2 PMDs should be configured for the vswitch when using multiqueue.
> +A least two PMDs should be configured for the vswitch when using multiqueue.
>  Using a single PMD will cause traffic to be enqueued to the same vhost queue
>  rather than being distributed among different vhost queues for a vhost-user
>  interface.
>  
>  If traffic destined for a VM configured with multiqueue arrives to the vswitch
> -via a physical DPDK port, then the number of rxqs should also be set to at
> -least 2 for that physical DPDK port. This is required to increase the
> +via a physical DPDK port, then the number of Rx queues should also be set to at
> +least two for that physical DPDK port. This is required to increase the
>  probability that a different PMD will handle the multiqueue transmission to the
>  guest using a different vhost queue.
>  
> 



More information about the dev mailing list