fields are populated with default values before the dev
init.
Also to avoid read error on unsupported ports, the flow control parameters
are now read only when user is trying to configure/update it.
Signed-off-by: Sugesh Chandran
---
lib/netdev-dpdk.c | 13 +
1 file changed, 5 insertions(+),
fields are populated with default values before the dev
init.
Also to avoid read error on unsupported ports, the flow control parameters
are now read only when user is trying to configure/update it.
Signed-off-by: Sugesh Chandran
---
lib/netdev-dpdk.c | 13 ++---
1 file changed, 6 insertions(+),
error on unsupported ports, the flow control parameters
are now read only when user is trying to configure/update it.
Signed-off-by: Sugesh Chandran
---
V1 -> V2
Read DPDK port flow-control parameters only when reconfiguration is required.
This will avoid flow control read error on unsupported
s now validating all the 'rte_eth_fc_conf' fields before
trying to configuring the dpdk ethdev. Hence OVS can no longer set the
'dont care' fields to just '0' as before. This commit make sure all the
'rte_eth_fc_conf' fields are populated with default values before the dev
init.
Signed-off-by: Sug
fields are populated with default values before the dev
init.
Signed-off-by: Sugesh Chandran
---
lib/netdev-dpdk.c | 15 ---
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index 5310373..58f77d2 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netd
fields are populated with default values before the dev
init.
Signed-off-by: Sugesh Chandran
---
lib/netdev-dpdk.c | 15 ---
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index d82c9e2..18f94c4 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netd
s now validating all the 'rte_eth_fc_conf' fields before
trying to configuring the dpdk ethdev. Hence OVS can no longer set the
'dont care' fields to just '0' as before. This commit make sure all the
'rte_eth_fc_conf' fields are populated with default values before the dev
init.
Signed-off-by: Suges
ot;:06:00.1", the
configuration option in OVS would be
ovs-vsctl set Open_vSwitch .
other_config:dpdk-whitelist-pci-ids=":06:00.0,:06:00.1"
To update the whitelist ports, OVS daemon has to be restarted.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.co
with
prefix 'ovs-1234A-' in the filesystem.
The following configuration option is used to enable the feature and changing
this value requires restarting the daemon.
ovs-vsctl --no-wait set Open_vSwitch . other_config:dpdk-isolate-mem=true
Signed-off-by: Sugesh Chandran <sugesh.ch
drivers at the time
of vswitchd init.
Sugesh Chandran (2):
Adding DPDK configuration option to isolate rte-mempool allocation.
Adding configuration option to whitelist DPDK physical ports.
lib/dpdk.c | 50 ++
vswitchd/vswitch.xml | 44
ot;:06:00.1", the
configuration option in OVS would be
ovs-vsctl set Open_vSwitch .
other_config:dpdk-whitelist-pci-ids=":06:00.0,:06:00.1"
To update the whitelist ports, OVS daemon has to be restarted.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.co
with
prefix 'ovs-12345' in the filesystem.
The following configuration option is used to enable the feature and changing
this value requires restarting the daemon.
ovs-vsctl --no-wait set Open_vSwitch . other_config:dpdk-isolate-mem=true
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.
drivers at the time
of vswitchd init.
Sugesh Chandran (2):
Adding DPDK configuration option to isolate rte-mempool allocation.
Adding configuration option to whitelist DPDK physical ports.
lib/daemon.c | 9
lib/daemon.h | 1 +
lib/dpdk.c | 63
cessing cost per TX packets in nsec : 106.53
The offered performance gain is approx 30%.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.
is also extended with new CLONE action to define
the packet cloning when the actions are combined. The lenght in the CLONE
action specifies the size of nested action set.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsso
-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Acked-by: Joe Stringer <j...@ovn.org>
---
ofproto/ofproto-dpif-xlate.c | 1 +
1 file changed, 1 insertion(
igned-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
---
lib/netdev-native-tnl.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/netdev-native-tnl.c b/li
igned-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
---
lib/netdev-native-tnl.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/netdev-native-tnl.c b/li
-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Acked-by: Joe Stringer <j...@ovn.org>
---
ofproto/ofproto-dpif-xlate.c | 1 +
1 file changed, 1 insertion(
Avoiding checksum validation in conntrack module if it is already verified
in DPDK physical NIC ports.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Co-authored-by: Darrell Ball <db...@vmware.com>
Signed-off-by: Darrell Ball <db...@vmware.com>
Acked-by
rx checksum
offload in conntrack module.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Co-authored-by: Darrell Ball <db...@vmware.com>
Signed-off-by: Darrell Ball <db...@vmware.com>
Acked-by: Antonio Fishetti <antonio.fische...@intel.com>
Acked-by: Darrel
v1->v2
- Rebased on master
- Added acked-by and tested-by tags in commit message
Sugesh Chandran (2):
dp-packet : Update DPDK rx checksum validation functions.
conntrack : Use Rx checksum offload feature on DPDK ports for
conntrack.
Avoiding checksum validation in conntrack module if it is already verified
in DPDK physical NIC ports.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Co-authored-by: Darrell Ball <db...@vmware.com>
Signed-off-by: Darrell Ball <db...@vmware.com>
---
li
rx checksum
offload in conntrack module.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Co-authored-by: Darrell Ball <db...@vmware.com>
Signed-off-by: Darrell Ball <db...@vmware.com>
---
lib/dp-packet.h | 28 ++--
1 file changed, 26 insertio
ned-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Co-authored-by: Darrell Ball <db...@vmware.com>
Signed-off-by: Darrell Ball <db...@vmware.com>
Sugesh Chandran (2):
dp-packet : Update DPDK rx checksum validation functions.
conntrack : Use Rx checksum offload feature on
Avoiding checksum validation in conntrack module if it is already verified
in DPDK physical NIC ports.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/conntrack.c | 63 -
lib/dp-packet.h | 22
2
DPDK ports use masks while reporting rx checksum flags. OVS should use these
mask along with reported checksum flag while validating the rx checksum.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Suggested-by: Darrell Ball <db...@vmware.com>
---
lib/dp-packet.h | 6 ++
This patch set removes the recirculation of encapsulated tunnel packets
if possible. It is done by computing the post tunnel actions at the time of
translation. The combined nested action set are programmed in the datapath
using CLONE action.
Signed-off-by: Sugesh Chandran <sugesh.ch
-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
---
ofproto/ofproto-dpif-xlate.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/ofproto/ofproto-dpif-xlate.c
is also extended with new CLONE action to define
the packet cloning when the actions are combined. The lenght in the CLONE
action specifies the size of nested action set.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsso
Outputting to a patch port is translated by its peer patch port actions.
Refactoring the translation part to use later in the patch series for the
tunnel push.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Avoiding checksum validation in conntrack module if it is already verified
in DPDK physical NIC ports.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/conntrack.c | 71 ++---
lib/dp-packet.h | 22 ++
2
DPDK ports use masks while reporting rx checksum flags. OVS should use these
mask along with reported checksum flag while validating the rx checksum.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Suggested-by: Darrell Ball <db...@vmware.com>
---
lib/dp-packet.h | 6 ++
acked-by and tested-by tags in commit message
Sugesh Chandran (2):
dp-packet : Update the rx checksum validation to use DPDK checksum
masks.
conntrack : Use Rx checksum offload feature on DPDK ports for
conntrack.
lib/conntrack.c | 71 ++---
This patch set removes the recirculation of encapsulated tunnel packets
if possible. It is done by computing the post tunnel actions at the time of
translation. The combined nested action set are programmed in the datapath
using CLONE action.
Signed-off-by: Sugesh Chandran <sugesh.ch
-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
---
ofproto/ofproto-dpif-xlate.c | 73
1 file changed, 73 inserti
Functions to update the xlate context with relevant tunnel fields to continue
the following translation on tunneled packets.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh
Outputting to a patch port is translated by its peer patch port actions.
Refactoring the translation part to use later in the patch series for the
tunnel push.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
logic is also extended with new CLONE action to define
the packet cloning when the actions are combined. The lenght in the CLONE
action specifies the size of nested action set.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsso
queue-id: 0
polling cycles(% of total polling cycles):273103986760 (73.98%)
processing cycles(% of total processing cycles):401626635996
(35.93%)
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/dpif-netdev.c
to know how useful stats at queue granularity.
Sugesh Chandran (1):
dpif-netdev : Include Rxq processing cycles information in the queue
stats
lib/dpif-netdev.c | 155 +-
1 file changed, 131 insertions(+), 24 deletions(-)
--
2.7.4
Avoiding checksum validation in conntrack module if it is already verified
in DPDK physical NIC ports.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/conntrack.c | 58 -
1 file changed, 37 insertions(+), 21 del
action with length to mark the tunnel combine action
set.
- Update the datapath trace display functions to handle CLONE.
- Fixed test cases to work with CLONE action.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
is using the new CLONE action to define
the packet cloning when the actions are combined. The lenght in the CLONE
action specifies the size of nested action set.
It also fixing the test suites failures that are introduced by nested CLONE
action in tunneling.
Signed-off-by: Sugesh Chandran
-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
---
lib/dpif-netdev.c | 18 +--
ofproto/ofproto-dpif-xlate.c | 271 ++-
Clone action is optimized by cloning a batch of packets together instead of
executing independently on every packet in a batch.
Rebased onto commit: edb417cc256e7d5f02f230867865c8cf7f26e542
ovs-ofctl: fix memory leak reported by valgrind.
Signed-off-by: Sugesh Chandran
the tunnel combine action
set.
- Update the datapath trace display functions to handle CLONE.
- Fixed test cases to work with CLONE action.
Sugesh Chandran (2):
odp-execute : Apply clone action on batch of packets instead of one by
one.
tunneling: Avoid recirculation on datapath by computing
too.
Rx checksum offloading still offers 8-9% of improvement on VxLAN tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Acked-by: Jesse Gross <je...@kernel.org>
---
v9
- Update
. These dummy new actions used as a delimiters
for combined action subset.
In our testing this patch improves the VxLAN encapsulation performance upto 30%.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Signed-off-by: Zoltán Balogh <zoltan.bal...@ericsson.com>
Co-authored-by: Z
too.
Rx checksum offloading still offers 8-9% of improvement on VxLAN tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Acked-by: Jesse Gross <je...@kernel.org>
---
v8
- Upda
too.
Rx checksum offloading still offers 8-9% of improvement on VxLAN tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Acked-by: Jesse Gross <je...@kernel.org>
---
v8
- Upda
too.
Rx checksum offloading still offers 8-9% of improvement on VxLAN tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Acked-by: Jesse Gross <je...@kernel.org>
---
in this patch. Can add when the
flow insert approach is finalized.
3) hardware specific rules are populated only at emc level. dp_classifier also
need to be populated accordingly.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
include/openvswitch/flow.h| 9 +-
include/openv
t processing stages.
3) Can work with P4
Sugesh Chandran (1):
Pipeline packet processing in OVS using FVL flow director.
include/openvswitch/flow.h| 9 +-
include/openvswitch/packets.h | 15 ++-
lib
.
Rx checksum offloading still offers 8-9% of improvement on VxLAN tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
Acked-by: Jesse Gross <je...@kernel.org>
---
v6
- Avoid
tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
v6
- Avoid unnecessary reconfiguration of DPDK ports when Rx checksum offload
support is unavailable.
- Limit the 'Rx checksum o
tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
v5
- Reset the checksum flag at common tunnel pop function than in
'udp_extract_tnl_md' function.
v4
- Unconditonally cle
tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
v4
- Unconditonally clear off the checksum flag one time in pop operation than
doing
separately in IP and UDP layers.
v3
-
tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
v4
- Unconditonally clear off the checksum flag one time in pop operation than
doing
separately in IP and UDP layers.
v3
-
tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
v3
- Reset the checksum offload flags in tunnel pop operation after the validation.
- Reconfigure the dpdk port with rx ch
tunneling
decapsulation even though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
v2
- Set Rx checksum enabled by default.
- Modified commit message, explaining the tradeoff with tx checksum offloading.
- Use dpd
though the SSE vector Rx function is disabled in DPDK poll
mode driver.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/dp-packet.h | 8
lib/dpif-netdev.c | 33
lib/netdev-dpdk.c | 103 ---
Interface dpdk0 options:tx-flow-ctrl=true'
The flow control parameters can be turned off by setting 'false' to the
respective parameter. To dsiable the flow control at tx side,
'ovs-vsctl set Interface dpdk0 options:tx-flow-ctrl=false'
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.
Interface dpdk0 options:tx-flow-ctrl=on'
The flow control parameters can be turned off by setting 'off' to the
respective parameter. To turn off the flow control at tx side,
'ovs-vsctl set Interface dpdk0 options:tx-flow-ctrl=off'
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
--
drop in
PHY-PHY case due to the overhead of validating invalid checksum flag reported
by NIC.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/dpif-netdev.c | 40
lib/netdev-dpdk.c | 39 +-
Optimizing VxLAN tunneling performance in userspace datapath by offloading
the rx checksum validation on tunnel packets to the NIC when it is supported.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/dpif-netdev.c| 73 --
lib/
Optimizing vxlan tunneling performance in userspace datapath using
flow director feature in Fortville NIC DPDK ports. OVS uses metadata
reported by NIC to improve the flow lookup performance on VxLAN
packets.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/autom
pport to netdev_tunnel_config")
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/flow.c | 7 ++---
lib/match.c | 12
lib/meta-flow.c | 10 +--
lib/netdev-vport.c | 10 +++
lib/odp-util.c
pport to netdev_tunnel_config")
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/flow.c | 7 ++---
lib/match.c | 12
lib/meta-flow.c | 10 +--
lib/netdev-vport.c | 10 +++
lib/odp-util.c
Adding a new field called protocol in flow tunnel structure to verify the
validity
of tunnel metadata. This field avoids the need of resetting and validating the
entire ipv4/ipv6 tunnel destination address which caused a serious performance
drop.
Signed-off-by: Sugesh Chandran <sugesh.ch
Adding a new field called protocol in flow tunnel structure to verify the
validity
of tunnel metadata. This field avoids the need of resetting and validating the
entire ipv4/ipv6 tunnel destination address which caused a serious performance
drop.
Signed-off-by: Sugesh Chandran <sugesh.ch
Adding a new flag to verify the validity of tunnel metadata. This flag avoids
the need of resetting and validating the entire ipv4/ipv6 tunnel destination
address which caused a serious performance drop.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/
Adding a new flag to validate if the tunnel metadata is valid/not. This flag
avoids
the need of resetting and validating the entire ipv4/ipv6 tunnel destination
address which caused a serious performance drop.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
lib/match.c
Adding a self-guide for configuring native userspace tunneling in OVS
with/without DPDK ports. This document also provides necessary debugging
commands to identify and resolve the userspace tunneling issues in OVS.
Signed-off-by: Sugesh Chandran <sugesh.chand...@intel.com>
---
README-
for populating the host information in OVN Southbound DB.
Signed-off-by: Sugesh Chandran sugesh.chand...@intel.com
---
ovn/controller/automake.mk | 2 ++
ovn/controller/ovn-controller.h | 4 +++-
ovn/controller/pn_sysinfo.c | 46 +
ovn/controller
From: sugeshch sugesh.chand...@intel.com
This patch adds the physical bridge information in ovn-controller man page. The
bridge is used for userspace tunneling.
Signed-off-by: Sugesh Chandran sugesh.chand...@intel.com
---
ovn/controller/ovn-controller.8.xml | 11 +++
1 file changed, 11
This is the patchset for making necessary OVN schema changes for enabling DPDK
ports and userspace tunneling in OVN-Openstack deployment. This patchset
creates tables to expose relevant Openvswitch table information to Openstack
Neutron via OVN Southbound and Northbound DBs. This physical host
From: sugeshch sugesh.chand...@intel.com
This patch contains changes for OVN southbound DB schema to expose Openvswitch/
hypervisor details. A new table named PN_sysinfo is created in the OVN
Southbound DB to store the information of each chassis.
Signed-off-by: Sugesh Chandran sugesh.chand
From: sugeshch sugesh.chand...@intel.com
This patch contains changes for OVN northbound DB schema to expose Openvswitch/
hypervisor details to CMS. A new table named 'Sys_Info is created in the OVN
Northbound DB. Each row in the table holds information about a chassis.
Signed-off-by: Sugesh
79 matches
Mail list logo