[dpdk-dev] [PATCH] doc: add basic howto for flow API

Ori Kam orika at mellanox.com
Thu Oct 19 08:55:39 CEST 2017


As the rte flow is a new complex module in the DPDK.
In order to ease developers in to using this feature
it was suggested to supply a simple howto doc.

Signed-off-by: Ori Kam <orika at mellanox.com>
---
 doc/guides/howto/index.rst    |    1 +
 doc/guides/howto/rte_flow.rst |  307 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 308 insertions(+), 0 deletions(-)
 create mode 100644 doc/guides/howto/rte_flow.rst

diff --git a/doc/guides/howto/index.rst b/doc/guides/howto/index.rst
index 468939b..0a22737 100644
--- a/doc/guides/howto/index.rst
+++ b/doc/guides/howto/index.rst
@@ -43,3 +43,4 @@ HowTo Guides
     virtio_user_for_container_networking
     virtio_user_as_exceptional_path
     packet_capture_framework
+    rte_flow
diff --git a/doc/guides/howto/rte_flow.rst b/doc/guides/howto/rte_flow.rst
new file mode 100644
index 0000000..02e944c
--- /dev/null
+++ b/doc/guides/howto/rte_flow.rst
@@ -0,0 +1,307 @@
+..  BSD LICENSE
+    Copyright(c) 2017 Mellanox Corporation. All rights reserved.
+    All rights reserved.
+
+    Redistribution and use in source and binary forms, with or without
+    modification, are permitted provided that the following conditions
+    are met:
+
+    * Redistributions of source code must retain the above copyright
+    notice, this list of conditions and the following disclaimer.
+    * Redistributions in binary form must reproduce the above copyright
+    notice, this list of conditions and the following disclaimer in
+    the documentation and/or other materials provided with the
+    distribution.
+    * Neither the name of Mellanox Corporation nor the names of its
+    contributors may be used to endorse or promote products derived
+    from this software without specific prior written permission.
+
+    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+    "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+    A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+    OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+    SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+    LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+    DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+    THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+    (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+    OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+
+Generic flow API - examples
+===========================
+
+This document function is to supply some concrete examples for programming
+flow rules.
+
+* Detail API can be found in the following link: :ref:`Generic flow API <Generic_flow_API>` .
+
+* Detail TestPMD command to set the flow rules can be found in the
+  following link: :ref:`TestPMD Flow rules <testpmd_rte_flow>`
+
+Simple IPv4 drop
+----------------
+
+Description
+~~~~~~~~~~~
+
+In this example we will create a simple rule that drops packets that their IPv4
+destination equals 192.168.3.2
+This code implements the following testpmd command::
+
+  tpmd> flow create 0 ingress pattern eth / vlan / ipv4 dst is 192.168.3.2 / end actions drop / end
+
+Code
+~~~~
+
+.. code-block:: c
+
+  /* create the attribute structure */
+  struct rte_flow_attr attr = {.ingress = 1};
+  struct rte_flow_item pattern[MAX_PATTERN_IN_FLOW];
+  struct rte_flow_action actions[MAX_ACTIONS_IN_FLOW];
+  struct rte_flow_item_etc eth;
+  struct rte_flow_item_vlan vlan;
+  struct rte_flow_item_ipv4 ipv4;
+  struct rte_flow *flow;
+  struct rte_flow_error error;
+
+  /* setting the eth to pass all packets */
+  pattern[0].type = RTE_FLOW_ITEM_TYPE_ETH;
+  pattern[0].spec = ð
+
+  /* set the vlan to pas all packets */
+  pattern[1] = RTE_FLOW_ITEM_TYPE_VLAN;
+  pattern[1].spec = &vlan;
+
+  /* set the dst ipv4 packet to the required value */
+  ipv4.hdr.dst_addr = htonl(0xc0a80302);
+  pattern[2].type = RTE_FLOW_ITEM_TYPE_IPV4;
+  pattern[2].spec = &ipv4;
+
+  /* end the pattern array */
+  pattern[3].type = RTE_FLOW_ITEM)TYPE_END;
+
+  /* create the drop action */
+  actions[0].type = RTE_FLOW_ACTION_TYPE_DROP;
+  actions[1].type = RTE_FLOW_ACTION_TYPE_END;
+
+  /* validate and create the flow rule */
+  if (!rte_flow_validate(port_id, &attr, pattern, actions, &error)
+      flow = rte_flow_create(port_id, &attr, pattern, actions, &error)
+
+Outputs sample
+~~~~~~~~~~~~~~
+terminal 1: running sample app flow rule disabled::
+
+  ./filter-program disable
+  [waiting for packets]
+
+terminal 2: running  scapy::
+
+  $scapy
+  welcome to Scapy
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.4', dst='192.168.3.1'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.5', dst='192.168.3.2'), iface = <some interface>, count 1)
+
+terminal 1: output log::
+
+  received packet with src ip = 176.80.50.4
+  received packet with src ip = 176.80.50.5
+
+terminal 1: running sample app flow rule enabled::
+
+  ./filter-program enabled
+  [waiting for packets]
+
+terminal 2: running  scapy::
+
+  $scapy
+  welcome to Scapy
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.4', dst='192.168.3.1'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.5', dst ='192.168.3.2'), iface = <some interface>, count 1)
+
+terminal 1: output log::
+
+  received packet with src ip = 176.80.50.4
+
+Range IPv4 drop
+----------------
+
+Description
+~~~~~~~~~~~
+
+In this example we will create a simple rule that drops packets that their IPv4
+destination is in the range of  192.168.3.0 to 192.168.3.255 This is done using
+mask.
+
+This code implements the following testpmd command::
+
+  tpmd> flow create 0 ingress pattern eth / vlan / ipv4 dst spec 192.168.3.0 dst mask 255.255.255.0 / end actions drop / end
+
+Code
+~~~~
+
+.. code-block:: c
+
+  struct rte_flow_attr attr = {.ingress = 1};
+  struct rte_flow_item pattern[MAX_PATTERN_IN_FLOW];
+  struct rte_flow_action actions[MAX_ACTIONS_IN_FLOW];
+  struct rte_flow_item_etc eth;
+  struct rte_flow_item_vlan vlan;
+  struct rte_flow_item_ipv4 ipv4;
+  struct rte_flow_item_ipv4 ipv4_mask;
+  struct rte_flow *flow;
+  struct rte_flow_error error;
+
+  /* setting the eth to pass all packets */
+  pattern[0].type = RTE_FLOW_ITEM_TYPE_ETH;
+  pattern[0].spec = ð
+
+  /* set the vlan to pas all packets */
+  pattern[1] = RTE_FLOW_ITEM_TYPE_VLAN;
+  pattern[1].spec = &vlan;
+
+  /* set the dst ipv4 packet to the required value */
+  ipv4.hdr.dst_addr = htonl(0xc0a80300);
+  ipv4_mask.hdr.dst_addr = htonl(0xffffff00);
+  pattern[2].type = RTE_FLOW_ITEM_TYPE_IPV4;
+  pattern[2].spec = &ipv4;
+  pattern[2].mask = &ipv4_mask;
+
+  /* end the pattern array */
+  pattern[3].type = RTE_FLOW_ITEM)TYPE_END;
+
+  /* create the drop action */
+  actions[0].type = RTE_FLOW_ACTION_TYPE_DROP;
+  actions[1].type = RTE_FLOW_ACTION_TYPE_END;
+
+  /* validate and create the flow rule */
+  if (!rte_flow_validate(port_id, &attr, pattern, actions, &error)
+      flow = rte_flow_create(port_id, &attr, pattern, actions, &error)
+
+Outputs sample
+~~~~~~~~~~~~~~
+terminal 1: running sample app flow rule disabled::
+
+  ./filter-program disable
+  [waiting for packets]
+
+terminal 2: running  scapy::
+
+  $scapy
+  welcome to Scapy
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.4', dst='192.168.3.1'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.5', dst='192.168.3.2'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.6', dst='192.168.5.2'), iface = <some interface>, count 1)
+
+terminal 1: output log::
+
+  received packet with src ip = 176.80.50.4
+  received packet with src ip = 176.80.50.5
+  received packet with src ip = 176.80.50.6
+
+terminal 1: running sample app flow rule enabled::
+
+  ./filter-program enabled
+  [waiting for packets]
+
+terminal 2: running  scapy::
+
+  $scapy
+  welcome to Scapy
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.4', dst='192.168.3.1'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.5', dst='192.168.3.2'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q()/IP(src='176.80.50.6', dst='192.168.5.2'), iface = <some interface>, count 1)
+
+terminal 1: output log::
+
+  received packet with src ip = 176.80.50.6
+
+Send vlan to queue
+------------------
+
+Description
+~~~~~~~~~~~
+
+In this example we will create a rule that routes all vlan id 123 to queue 3.
+
+This code implements the following testpmd command::
+
+  tpmd> flow create 0 ingress pattern eth / vlan vid spec 123 / end actions queue index 3 / end
+
+Code
+~~~~
+
+.. code-block:: c
+
+  struct rte_flow_attr attr = {.ingress = 1};
+  struct rte_flow_item pattern[MAX_PATTERN_IN_FLOW];
+  struct rte_flow_action actions[MAX_ACTIONS_IN_FLOW];
+  struct rte_flow_item_etc eth;
+  struct rte_flow_item_vlan vlan;
+  struct rte_flow_action_queue queue = { .index = 3 };
+  struct rte_flow *flow;
+  struct rte_flow_error error;
+
+  /* setting the eth to pass all packets */
+  pattern[0].type = RTE_FLOW_ITEM_TYPE_ETH;
+  pattern[0].spec = ð
+
+  /* set the vlan to pas all packets */
+  vlan.vid = 123;
+  pattern[1] = RTE_FLOW_ITEM_TYPE_VLAN;
+  pattern[1].spec = &vlan;
+
+  /* end the pattern array */
+  pattern[2].type = RTE_FLOW_ITEM)TYPE_END;
+
+  /* create the drop action */
+  actions[0].type = RTE_FLOW_ACTION_TYPE_QUEUE;
+  actions[0].conf = &queue
+  actions[1].type = RTE_FLOW_ACTION_TYPE_END;
+
+  /* validate and create the flow rule */
+  if (!rte_flow_validate(port_id, &attr, pattern, actions, &error)
+      flow = rte_flow_create(port_id, &attr, pattern, actions, &error)
+
+Outputs sample
+~~~~~~~~~~~~~~
+terminal 1: running sample app flow rule disabled::
+
+  ./filter-program disable
+  [waiting for packets]
+
+terminal 2: running  scapy::
+
+  $scapy
+  welcome to Scapy
+  >> sendp(Ether()/Dot1Q(vlan=123)/IP(src='176.80.50.4', dst='192.168.3.1'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q(vlan=50)/IP(src='176.80.50.5', dst='192.168.3.2'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q(vlan=123)/IP(src='176.80.50.6', dst='192.168.5.2'), iface = <some interface>, count 1)
+
+terminal 1: output log::
+
+  received packet with src ip = 176.80.50.4 sent to queue 2
+  received packet with src ip = 176.80.50.5 sent to queue 1
+  received packet with src ip = 176.80.50.6 sent to queue 0
+
+terminal 1: running sample app flow rule enabled::
+
+  ./filter-program enabled
+  [waiting for packets]
+
+terminal 2: running  scapy::
+
+  $scapy
+  welcome to Scapy
+  >> sendp(Ether()/Dot1Q(vlan=123)/IP(src='176.80.50.4', dst='192.168.3.1'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q(vlan=50)/IP(src='176.80.50.5', dst='192.168.3.2'), iface = <some interface>, count 1)
+  >> sendp(Ether()/Dot1Q(vlan=123)/IP(src='176.80.50.6', dst='192.168.5.2'), iface = <some interface>, count 1)
+
+terminal 1: output log::
+
+  received packet with src ip = 176.80.50.4 sent to queue 3
+  received packet with src ip = 176.80.50.5 sent to queue 1
+  received packet with src ip = 176.80.50.6 sent to queue 3
-- 
1.7.1



More information about the dev mailing list