From: wenxu
In case almost or all available ports are taken, clash resolution can
take a very long time, resulting in pmd hang in conntrack.
This can happen when many to-be-natted hosts connect to same
destination:port (e.g. a proxy) and all connections pass the same SNAT.
Pick a random offset
From: wenxu
This commit splits the nested loop used to search the unique ports for
the reverse tuple.
It affects only the dnat action, giving more precedence to the dnat
range, similarly to the kernel dp, instead of searching through the
default ephemeral source range for each destination port.
From: wenxu
Like the kernel datapath. The sport nat range for well-konwn origin
sport should limit in the well-known ports.
Signed-off-by: wenxu
Acked-by: Paolo Valerio
---
lib/conntrack.c | 12 ++--
1 file changed, 10 insertions(+), 2 deletions(-)
diff --git a/lib/conntrack.c b/lib/
>Problem Statement:
>OVS flushes and subsequently repopulates its route
>cache whenever it receives a netlink notification
>about kernel interface change. At the same time the
>port addition triggers a revalidation of all
>datapath flow cache entries. The revalidation of
>egress tunnel flows depend
From: Paolo Valerio
Date: 2022-01-12 18:19:25
To: we...@ucloud.cn,i.maxim...@ovn.org
Cc: d...@openvswitch.org
Subject: Re: [PATCH v8 3/3] conntrack: limit port clash resolution
attempts>Hello wenxu,
>
>I tested a bit more the patch, and it seems to effectively limit the
>number of atte
> From: Kumar Amber
>
> This patch adds error checking of packet hashes to the mfex
> autovalidator infrastructure, ensuring that hashes calculated by
> optimized mfex implementations is identical to the scalar code.
>
> This patch avoids calculating the software hash of the packet again
> if th
OVN LTS releases have a lot of ambiguity, so this is intended to codify
LTS support and cadence.
Signed-off-by: Mark Michelson
---
Documentation/internals/release-process.rst | 28 +
1 file changed, 18 insertions(+), 10 deletions(-)
diff --git a/Documentation/internals/relea
> This commit adds the AVX512 implementation of the push_vlan action.
> The implementation here is auto-validated by the miniflow
> extract autovalidator, hence its correctness can be easily
> tested and verified.
>
> Signed-off-by: Emma Finn
Hi Emma, thanks for the patch.
I think given the con
> This commit adds the AVX512 implementation of the pop_vlan action.
> The implementation here is auto-validated by the miniflow
> extract autovalidator, hence its correctness can be easily
> tested and verified.
>
> Signed-off-by: Emma Finn
Hi Emma, some comments below.
> ---
> lib/odp-execu
> -Original Message-
> From: Finn, Emma
> Sent: Wednesday, January 12, 2022 9:43 AM
> To: d...@openvswitch.org; Van Haaren, Harry ;
> Amber, Kumar ; Stokes, Ian ;
> i.maxim...@ovn.org
> Cc: Finn, Emma
> Subject: [PATCH v5 6/8] odp-execute: Add ISA implementation of actions.
>
> This c
> This commit adds a new command to allow the user to switch
> the active action implementation at runtime. A probe function
> is executed before switching the implementation, to ensure
> the CPU is capable of running the ISA required.
>
> Usage:
> $ ovs-appctl dpif-netdev/action-impl-set scalar
> -Original Message-
> From: David Marchand
> Sent: Wednesday 12 January 2022 17:05
> To: Ferriter, Cian
> Cc: Phelan, Michael ; d...@openvswitch.org; Ilya
> Maximets
> Subject: Re: [ovs-dev] OVS DPDK MFEX Unit Tests Failing
>
> On Wed, Jan 12, 2022 at 5:54 PM Ferriter, Cian
> wrot
On 1/6/22 14:11, Van Haaren, Harry wrote:
>> -Original Message-
>> From: Finn, Emma
>> Sent: Wednesday, January 5, 2022 4:54 PM
>> To: d...@openvswitch.org; Van Haaren, Harry ;
>> Amber, Kumar
>> Cc: Finn, Emma
>> Subject: [PATCH v4 0/9] Actions Infrastructure + Optimizations
>>
>> ---
>
On Wed, Jan 12, 2022 at 5:54 PM Ferriter, Cian wrote:
> I tested your fix and it works, but I had to modify the port number from 0 to
> 1. If I leave it at 0, the tests still fail.
It depends on what DPDK ports get initialised on your system.
On mine, I had no pci device bound to vfio-pci (and u
Hi David,
I tested your fix and it works, but I had to modify the port number from 0 to
1. If I leave it at 0, the tests still fail.
I put the modifications I made inline below.
Perhaps we need to wildcard this number?
Thanks,
Cian
> -Original Message-
> From: dev On Behalf Of David
On 1/12/22 17:31, David Marchand wrote:
> On Wed, Jan 12, 2022 at 5:17 PM David Marchand
> wrote:
>>
>> On Wed, Jan 12, 2022 at 4:48 PM Phelan, Michael
>> wrote:
>>>
>>> Hi,
>>>
>>> During internal testing of the AVX-512 CI, a bug was picked up on the OVS
>>> master branch which resulted in the
Hi,
During internal testing of the AVX-512 CI, a bug was picked up on the OVS
master branch which resulted in the MFEX unit tests consistently failing. I
believe the bug was introduced by commit
d446dcb7e03fc7bd4e3050c83c22233b0a46d364 "system-dpdk: Refactor common logs
matching". It looks like
From: Wilson Peng
CT marks which are loaded in non-first commit will be lost in ovs-windows.In
linux OVS,
the CT mark setting with same flow could be set successfully.
Currenlty Ovs-windows will create one new CT with the flowKey(Extracted from
the packet itself)
If the packet is already done
On Wed, Jan 12, 2022 at 5:17 PM David Marchand
wrote:
>
> On Wed, Jan 12, 2022 at 4:48 PM Phelan, Michael
> wrote:
> >
> > Hi,
> >
> > During internal testing of the AVX-512 CI, a bug was picked up on the OVS
> > master branch which resulted in the MFEX unit tests consistently failing. I
> > be
This commit fixes the minimum packet size for the vlan/ipv4/tcp
traffic profile, which was previously incorrectly set.
This commit also disallows any fragmented IPv4 packets from being
matched in the optimized miniflow-extract, avoiding complexity of
handling fragmented packets and using scalar fa
Bleep bloop. Greetings Harry van Haaren, I am a robot and I have tried out
your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Unexpected sign-offs from developers who are not authors or co-authors
or committ
On 12/4/21 00:06, Mike Pattrick wrote:
> Currently ingress policing uses the basic classifier to apply traffic
> control filters if hardware offload is not enabled, in which case it
> uses matchall. This change changes the behavior to always use matchall,
> and fall back onto basic if the kernel is
On Wed, Jan 12, 2022 at 4:48 PM Phelan, Michael
wrote:
>
> Hi,
>
> During internal testing of the AVX-512 CI, a bug was picked up on the OVS
> master branch which resulted in the MFEX unit tests consistently failing. I
> believe the bug was introduced by commit
> d446dcb7e03fc7bd4e3050c83c22233
From: Kumar Amber
This patch adds error checking of packet hashes to the mfex
autovalidator infrastructure, ensuring that hashes calculated by
optimized mfex implementations is identical to the scalar code.
This patch avoids calculating the software hash of the packet again
if the optimized mini
Following from the MFEX Optimizations IPv6 + Hashing patchset,
https://patchwork.ozlabs.org/project/openvswitch/list/?series=275590
This patchset introduces the optimization as described at OVS Conference;
https://www.openvswitch.org/support/ovscon2021/#T32
https://youtu.be/X_uPybauF3g?list=PLaJlR
From: Kumar Amber
This commit adds IPv4 profile specific hashing which
uses fixed offsets into the packet to improve hashing
performance.
Signed-off-by: Kumar Amber
Signed-off-by: Harry van Haaren
Co-authored-by: Harry van Haaren
---
v5:
- Rename "hash_len" to "key_len" to describe its use
> This commit introduced the auto-validation function which
> allows users to compare the batch of packets obtained from
> different action implementations against the linear
> action implementation.
>
> The autovalidator function can be triggered at runtime using the
> following command:
>
> $ o
On 12/24/21 15:07, Eelco Chaudron wrote:
+class OFPFlowFactory:
See my comments on patch 8, where I do think we should get rid of this class,
and update the OFPFlow class to take this string at init.
Being more OOO it would look like this:
def __init__(self, ofp_string, id=None):
"""Co
> -Original Message-
> From: Finn, Emma
> Sent: Wednesday, January 12, 2022 9:43 AM
> To: d...@openvswitch.org; Van Haaren, Harry ;
> Amber, Kumar ; Stokes, Ian ;
> i.maxim...@ovn.org
> Cc: Finn, Emma
> Subject: [PATCH v5 2/8] odp-execute: Add function pointer for pop_vlan action.
>
>
> -Original Message-
> From: Finn, Emma
> Sent: Wednesday, January 12, 2022 9:43 AM
> To: d...@openvswitch.org; Van Haaren, Harry ;
> Amber, Kumar ; Stokes, Ian ;
> i.maxim...@ovn.org
> Cc: Finn, Emma
> Subject: [PATCH v5 1/8] odp-execute: Add function pointers to odp-execute for
> dif
On Wed, Jan 12, 2022 at 8:44 AM Lorenzo Bianconi
wrote:
>
> > On Tue, Jan 11, 2022 at 2:19 PM Lorenzo Bianconi
> > wrote:
> > >
> > > Since commit 4597317f1 ("Introduce incremental processing for northd"),
> > > we rely on (partial) IP for ovn-northd. We need to track SB port_binding
> > > option
Hello Sunil, Marko and Ian.
Mike worked to identify the reason for the performance issue
reported by you a while ago. He summarized below. I wonder
if you can give a try on his patch too and tell us if we are
on the right track.
Thanks,
fbl
On Wed, Jan 05, 2022 at 03:01:47PM -0500, Mike Pattri
> On Tue, Jan 11, 2022 at 2:19 PM Lorenzo Bianconi
> wrote:
> >
> > Since commit 4597317f1 ("Introduce incremental processing for northd"),
> > we rely on (partial) IP for ovn-northd. We need to track SB port_binding
> > option column in order to notify the CMS whenever the controller
> > receives
Replace dibbler dhcp6 server with dhcpd since the former is no longer
maintained.
Signed-off-by: Lorenzo Bianconi
---
tests/atlocal.in| 4 ++--
tests/system-ovn.at | 34 +++---
2 files changed, 13 insertions(+), 25 deletions(-)
diff --git a/tests/atlocal.in b/te
On 12/17/21 15:37, Eelco Chaudron wrote:
On 22 Nov 2021, at 12:22, Adrian Moreno wrote:
Based on meta-field information extracted by extract_ofp_fields,
autogenerate the right decoder to be used
.
Signed-off-by: Adrian Moreno
---
build-aux/automake.mk| 3 +-
build-aux/g
On 1/6/22 12:45, Harry van Haaren wrote:
> This commit improves handling of packets where the allocated memory
> is less than 64 bytes. In the DPDK datapath this never matters, as
> an mbuf always pre-allocates enough space, however this can occur in
> test environments such as the dummy netdev.
T
On 12/17/21 14:53, Eelco Chaudron wrote:
On 22 Nov 2021, at 12:22, Adrian Moreno wrote:
In order to be able to reuse the core extaction logic, split the command
extaction -> extraction
in two parts. The core extraction logic is moved to python/build while
the command that writes the dif
On 12/17/21 14:41, Eelco Chaudron wrote:
See some minor comment below.
//Eelco
On 22 Nov 2021, at 12:22, Adrian Moreno wrote:
Some openflow or dpif flows encode their arguments in lists, eg:
"some_action(arg1,arg2,arg3)". In order to decode this in a way that can
be then stored and queried
Ilya Maximets writes:
> Since we're testing serialization, it also makes sense to test
> the opposite operation. Should be useful in the future for
> exploring possible optimizations.
>
> CMD: $ ./tests/ovstest json-string-benchmark
>
> Signed-off-by: Ilya Maximets
> ---
Acked-by: Aaron Conole
On 1/9/22 10:05, Sunil Pai G wrote:
> Currently, on failure to link with DPDK, the configure script provides
> an error message to update the PKG_CONFIG_PATH even though the cause of
> failure was missing dependencies. Improve the error message to include this
> scenario.
>
> Signed-off-by: Sunil
On 12/14/21 14:33, Kevin Traynor wrote:
> On 02/12/2021 21:16, David Marchand wrote:
>> When troubleshooting multiqueue setups, having per queue statistics helps
>> checking packets repartition in rx and tx queues.
>>
>> Per queue statistics are exported by most DPDK drivers (with capability
>> RTE
Hi,
Thanks for the reviewing.
Aaron Conole 于2022年1月12日周三 05:37写道：
> Peng He writes:
>
> > by using ipf_list's key instead of first frags' metadata can reduce
> > quite a lot of cache access as by the time calling ipf_ctx_eq, ipf_list
> > is cache-hot.
> >
> > Signed-off-by: Peng He
> > ---
>
On 1/11/22 23:54, Nick Bouliane wrote:
>> BTW, the up-to-date versions of some of these docs are available in
>> html form in the documentation:
>> https://docs.openvswitch.org/en/latest/ref/ovs-actions.7/
>
> oh, thanks ! any plan to continue generate the pdf/txt format for easy
> offline readi
we...@ucloud.cn writes:
> From: wenxu
>
> Splits the nested loop used to search the unique ports for the
> reverse tuple.
> It affects only the dnat action, giving more precedence to the dnat
> range, similarly to the kernel dp, instead of searching through the
> default ephemeral source range fo
Hello wenxu,
I tested a bit more the patch, and it seems to effectively limit the
number of attempts. There is a case with a sufficiently large port range
that will always tries the same ports.
E.g. (incresing the IPs you can reduce the port range):
actions=ct(commit,nat(dst=10.1.1.100-10.1.1.101
This commit adds the AVX512 implementation of the push_vlan action.
The implementation here is auto-validated by the miniflow
extract autovalidator, hence its correctness can be easily
tested and verified.
Signed-off-by: Emma Finn
---
lib/odp-execute-avx512.c | 62 ++
This commit adds the AVX512 implementation of the pop_vlan action.
The implementation here is auto-validated by the miniflow
extract autovalidator, hence its correctness can be easily
tested and verified.
Signed-off-by: Emma Finn
---
lib/odp-execute-avx512.c | 77 +++
This commit adds the AVX512 implementation of the action functionality.
Usage:
$ ovs-appctl dpif-netdev/action-impl-set avx512
Signed-off-by: Emma Finn
Acked-by: Harry van Haaren
---
Documentation/topics/dpdk/bridge.rst | 25 ++
Documentation/topics/testing.rst | 20 +---
NEW
From: Kumar Amber
This commit adds a new command to allow the user to enable the
actions autovalidator by default at build time thus allowing for
running unit test by default.
$ ./configure --enable-actions-default-autovalidator
Signed-off-by: Kumar Amber
Acked-by: Harry van Haaren
---
NEWS
This commit adds a new command to allow the user to switch
the active action implementation at runtime. A probe function
is executed before switching the implementation, to ensure
the CPU is capable of running the ISA required.
Usage:
$ ovs-appctl dpif-netdev/action-impl-set scalar
This commit
This commit introduced the auto-validation function which
allows users to compare the batch of packets obtained from
different action implementations against the linear
action implementation.
The autovalidator function can be triggered at runtime using the
following command:
$ ovs-appctl dpif-net
This commit introduces the initial infrastructure required to allow
different implementations for OvS actions. The patch introduces action
function pointers which allows user to switch between different action
implementations available. This will allow for more performance and flexibility
so the us
This commit removes the pop_vlan action from the large switch
and creates a separate function for batched processing. A function
pointer is also added to call the new batched function for the pop_vlan
action.
Signed-off-by: Emma Finn
Acked-by: Harry van Haaren
---
lib/odp-execute-private.c | 19
---
v5:
- Rebase to master
- Minor change to variable names
- Added Tags from Harry.
---
v4:
- Rebase to master
- Add ISA implementation of push_vlan action
---
v3:
- Refactored to fix unit test failures
- Removed some sign-off on commits
---
v2:
- Fix the CI build issues
---
This patchset introd
54 matches
Mail list logo
