From: Yinan <yinan.wang@intel.com>
To: dev@dpdk.org, maxime.coquelin@redhat.com, tiwei.bie@intel.com
Cc: zhihong.wang@intel.com, john.mcnamara@intel.com,
marko.kovacevic@intel.com, Yinan Wang <yinan.wang@intel.com>
Subject: [dpdk-dev] [PATCH] doc: clarify virtio PMD path selection
Date: Mon, 18 Nov 2019 05:29:23 -0500 [thread overview]
Message-ID: <20191118102923.9302-1-yinan.wang@intel.com> (raw)
From: Yinan Wang <yinan.wang@intel.com>
add virtio paths selection and usage introduction for better
virtio usability.
Signed-off-by: Yinan Wang <yinan.wang@intel.com>
---
.../virtio_paths_selection_and_usage.rst | 130 ++++++++++++++++++
1 file changed, 130 insertions(+)
create mode 100644 doc/guides/howto/virtio_paths_selection_and_usage.rst
diff --git a/doc/guides/howto/virtio_paths_selection_and_usage.rst b/doc/guides/howto/virtio_paths_selection_and_usage.rst
new file mode 100644
index 000000000..b5918eaf9
--- /dev/null
+++ b/doc/guides/howto/virtio_paths_selection_and_usage.rst
@@ -0,0 +1,130 @@
+.. SPDX-License-Identifier: BSD-3-Clause
+ Copyright(c) 2019 Intel Corporation.
+
+Virtio paths Selection and Usage
+================================
+
+Logically virtio-PMD has 9 paths based on the virtio features (Rx mergeable,In-order,Packed virtqueue)
+combinations, below are introduction of virtio three common features:
+
+* Rx mergeable: With this feature negotiated, device can receive larger packets by combining
+ individual descriptors.
+* In-order: Some devices always use descriptors in the same order in which they have been made
+ available, these devices can offer the VIRTIO_F_IN_ORDER feature. If this feature negotiated,
+ driver will use descriptors in order. Meanwhile, this knowledge allows device operate used ring
+ in batches and driver operate available ring in batches and such can decrease cache miss rate.
+* Packed virtqueue: The structure of packed virtqueue is different from split virtqueue,split
+ virtqueue is composed of available ring, used ring and descriptor table, while packed virtqueue
+ is composed of descriptor ring,driver event suppression and device event suppression. The idea
+ behind this is to improve performance by avoiding cache misses and and make it easier for devices
+ to implement.
+
+Virtio paths Selection
+----------------------
+
+If packed virtqueue is not negotiated, below split virtqueue paths can be selected
+according to below configuration:
+
+#. Split virtqueue mergeable path: If Rx mergeable is negotiated, in-order feature is
+ not negotiated, this path will be selected.
+#. Split virtqueue non-mergeable path: If Rx mergeable and in-order feature are not
+ negotiated, also Rx offload(s) are requested, this path can be selected.
+#. Split virtqueue in-order mergeable path: If in-order feature and Rx mergeable are
+ both negotiated, this path can be selected.
+#. Split virtqueue in-order non-mergeable path: If in-order feature is negotiated and
+ Rx mergeable is not negotiated, this path can be selected.
+#. Split virtqueue vectorized RX path: If Rx mergeable is disabled and no Rx offload
+ requested, this path can be selected.
+
+If packed virtqueue is negotiated, below packed virtqueue paths can be selected
+according to below configuration:
+
+#. Packed virtqueue mergeable path: If Rx mergeable is negotiated, in-order feature
+ is not negotiated, this path will be selected.
+#. Packed virtqueue non-mergeable path: If Rx mergeable and in-order feature are not
+ negotiated, also Rx offload(s) are requested, this path will be selected.
+#. Packed virtqueue in-order mergeable path: If in-order feature and Rx mergeable are
+ both negotiated, this path will be selected.
+#. Packed virtqueue in-order non-mergeable path: If in-order feature is negotiated and
+ Rx mergeable is not negotiated, this path will be selected.
+
+Rx/Tx callbacks of each Virtio path
+-----------------------------------
+
+Refer to above descriptions,virtio path and Rx/TX callbacks are auto selected by different parameters of
+vdev and workloads. Rx callbacks and Tx callbacks name for each Virtio Path are shown in following tables::
+
+ +----------------------------------------------------------------------------------------------------------+
+ | Virtio path | Rx callbacks | TX callbacks |
+ +----------------------------------------------------------------------------------------------------------+
+ |Split virtqueue mergeable path |virtio_recv_mergeable_pkts | virtio_xmit_pkts |
+ +----------------------------------------------------------------------------------------------------------+
+ |Split virtqueue non-mergeable path | virtio_recv_pkts | virtio_xmit_pkts |
+ +----------------------------------------------------------------------------------------------------------+
+ |Split virtqueue in-order mergeable path | virtio_recv_pkts_inorder | virtio_xmit_pkts_inorder|
+ +----------------------------------------------------------------------------------------------------------+
+ |Split virtqueue in-order non-mergeable path | virtio_recv_pkts_inorder | virtio_xmit_pkts_inorder|
+ +----------------------------------------------------------------------------------------------------------+
+ |Split virtqueue vectorized RX path | virtio_recv_pkts_vec | virtio_xmit_pkts |
+ +----------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue mergeable path | virtio_recv_mergeable_pkts_packed| virtio_xmit_pkts_packed |
+ +----------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue normal path | virtio_recv_pkts_packed | virtio_xmit_pkts_packed |
+ +----------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue in-order mergeable path | virtio_recv_mergeable_pkts_packed| virtio_xmit_pkts_packed |
+ +----------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue in-order normal path | virtio_recv_pkts_packed | virtio_xmit_pkts_packed |
+ +----------------------------------------------------------------------------------------------------------+
+
+Virtio paths Support Status from Release to Release
+---------------------------------------------------
+
+Virtio feature implementation:
+
+* In-order feature implemented in DPDK 18.08 by adding new Rx/TX callbacks
+ ``virtio_recv_pkts_inorder`` and ``virtio_xmit_pkts_inorder``.
+* Packed virtqueue implemented in DPDK 19.02 by adding new Rx/TX callbacks
+ ``virtio_recv_pkts_packed`` , ``virtio_recv_mergeable_pkts_packed`` and ``virtio_xmit_pkts_packed``.
+
+Virtio path number changes from release to release, all virtio paths support status are shown in below table::
+
+ +--------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Virtio path\ DPDK version | v16.11 | v17.02 | v17.05 | v17.08 | v17.11 | v18.02 | v18.05 | v18.08 | v18.11 | v19.02 | v19.05 | v19.08 |
+ +--------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Split virtqueue mergebale path | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y |
+ +--------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Split virtqueue non-mergeable path | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y |
+ +--------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Split virtqueue vectorized RX path | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y | Y |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Split virtqueue simple TX path | Y | Y | Y | Y | Y | Y | Y | N | N | N | N | N |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Split virtqueue in-order non-mergeable path | | | | | | | | Y | Y | Y | Y | Y |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Split virtqueue in-order mergeable path | | | | | | | | Y | Y | Y | Y | Y |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue mergeable path | | | | | | | | | | Y | Y | Y |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue non-mergeable path | | | | | | | | | | Y | Y | Y |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue in-order mergeable path | | | | | | | | | | Y | Y | Y |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+ |Packed virtqueue in-order non-mergeable path| | | | | | | | | | Y | Y | Y |
+ ---------------------------------------------------------------------------------------------------------------------------------------------------------+
+
+QEMU Support Status
+-------------------
+
+Qemu only support three path of Virtio-PMD: Split virtqueue mergebale path,
+Split virtqueue no-mergeable path,Split virtqueue vectorized RX path.
+
+How to Debug
+------------
+
+If you meet performance drop or some other issues after upgrading the driver
+or configuration, below steps can help you identify which path you selected and
+root cause faster.
+
+#. Run vhost/virtio test case;
+#. Run "perf top" and check virtio Rx/tx callback names;
+#. Identify which virtio path is selected refer to above table.
--
2.17.1
next reply other threads:[~2019-11-18 1:37 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-11-18 10:29 Yinan [this message]
2019-11-18 2:23 ` Stephen Hemminger
2019-11-25 15:49 ` Wang, Yinan
2019-11-18 9:55 ` Gavin Hu (Arm Technology China)
2019-11-26 16:08 ` [dpdk-dev] [PATCH v4] " Yinan
2019-11-27 2:43 ` Tiwei Bie
2019-11-26 21:30 ` [dpdk-dev] [PATCH] " Thomas Monjalon
2019-11-27 15:19 ` [dpdk-dev] [PATCH v5] " Yinan
2019-11-27 6:38 ` Tiwei Bie
2019-11-27 8:41 ` Maxime Coquelin
2019-11-28 0:14 ` Thomas Monjalon
2019-11-28 14:30 ` [dpdk-dev] [PATCH v6] " Maxime Coquelin
2019-11-28 18:17 ` Thomas Monjalon
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20191118102923.9302-1-yinan.wang@intel.com \
--to=yinan.wang@intel.com \
--cc=dev@dpdk.org \
--cc=john.mcnamara@intel.com \
--cc=marko.kovacevic@intel.com \
--cc=maxime.coquelin@redhat.com \
--cc=tiwei.bie@intel.com \
--cc=zhihong.wang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).