Hi Eelco,
MFEX v7 will be available shorty EOD today.
Some comments are inline.
> > + * for that packet.
> > + */
> > +uint32_t mfex_hit = (mf_mask & (1 << i));
>
> This was supposed to become a bool?
>
This cannot be a bool as this is used like a bit-mask and set
Hi Ilya,
How about this patch. Without this the fragment packet in the nat conntrack will
not work for the only first fragment do address nat.
BR
wenxu
From: we...@ucloud.cn
Date: 2021-06-18 14:45:50
To: i.maxim...@ovn.org
Cc: d...@openvswitch.org
Subject: [ovs-dev] [PATCH v2] ipf: fix only
Bleep bloop. Greetings Ihar Hrachyshka, 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.
git-am:
error: Failed to merge in the changes.
hint: Use 'git am --show-current-patch' to see the
Recently, we stopped leaking localport traffic through localnet ports
into fabric to avoid unnecessary flipping between chassis hosting the
same localport.
Despite the type name, in some scenarios localports are supposed to talk
outside the hosting chassis. Specifically, in OpenStack [1] metadata
Hi,
After the refactoring and rebasing, this patch doesn't seem
necessary anymore. I don't see value in keeping it.
Can we drop it? What do you think?
fbl
On Thu, Jul 01, 2021 at 04:06:10PM +0100, Cian Ferriter wrote:
> From: Harry van Haaren
>
> This commit moves the datapath lookup
On Sat, Jul 03, 2021 at 04:23:14PM +0800, wangyunjian wrote:
> From: Yunjian Wang
>
> This series include two fixes patches for ovs'comments
Thanks, applied.
___
dev mailing list
d...@openvswitch.org
The vswitchd database schema requires role names to be "master" or
"slave", but this code tried to use "primary" and "secondary".
Signed-off-by: Ben Pfaff
Reported-at: https://github.com/openvswitch/ovs-issues/issues/218
Fixes: 807152a4ddfb ("Use primary/secondary, not master/slave, as names for
On Thu, Jul 01, 2021 at 04:06:09PM +0100, Cian Ferriter wrote:
> From: Harry van Haaren
>
> Split the very large file dpif-netdev.c and the datastructures
> it contains into multiple header files. Each header file is
> responsible for the datastructures of that component.
>
> This logical split
On Tue, Jul 06, 2021 at 04:20:59PM -0300, Flavio Leitner wrote:
>
> Hi,
>
> I was reviewing the patch while testing and I can consistently
> loss 1Mpps (or more) on a P2P scenario with this flow table:
> ovs-ofctl add-flow br0 in_port=dpdk0,actions=output:dpdk1
>
> TX: 14Mpps
> RX without
Hi,
I was reviewing the patch while testing and I can consistently
loss 1Mpps (or more) on a P2P scenario with this flow table:
ovs-ofctl add-flow br0 in_port=dpdk0,actions=output:dpdk1
TX: 14Mpps
RX without patch: +12.6Mpps
RX with patch: 11.67Mpps
CPU: Intel(R) Xeon(R) Silver 4114 CPU @
On Tue, Jul 6, 2021 at 1:19 AM Frode Nordahl
wrote:
>
> On Mon, Jul 5, 2021 at 6:57 PM Numan Siddique wrote:
> >
> > On Mon, Jul 5, 2021 at 12:12 PM Frode Nordahl
> > wrote:
> > >
> > > On Wed, Jun 30, 2021 at 12:32 AM Numan Siddique
wrote:
> > > >
> > > > On Thu, Jun 10, 2021 at 10:13 AM
On 06/07/2021 03:51, li...@rc.inesa.com wrote:
>> On 05/07/2021 09:08, li...@rc.inesa.com wrote:
>>> Dear OVS Team,
>>>
>>> We are seeing an issue occurred in OVS version 2.11.0. Right now we are
>>> trying to encrypt our VXLAN packets in communication. The problem is, when
>>> the tunnel is
> -Original Message-
> From: Eelco Chaudron
> Sent: Tuesday, July 6, 2021 3:19 PM
> To: Ferriter, Cian
> Cc: ovs-dev@openvswitch.org; f...@sysclose.org; i.maxim...@ovn.org; Van
> Haaren,
> Harry ; Amber, Kumar ;
> Stokes, Ian
> Subject: Re: [v5 01/11] dpif-netdev: Add command line and
Hi Eelco ,
Here is the diff vor v6 vs v5 :
Patch 1 :
diff --git a/lib/dpif-netdev-private-extract.c
b/lib/dpif-netdev-private-extract.c
index 1aebf3656d..4987d628a4 100644
--- a/lib/dpif-netdev-private-extract.c
+++ b/lib/dpif-netdev-private-extract.c
@@ -233,7 +233,7 @@
Also, shorten the CONTROLLER_LOOP_STOPWATCH_NAME name as there is a bug
in lib/stopwatch.c which fails to report an error when the stopwatch
name is longer than 32 characters. CONTROLLER_LOOP_STOPWATCH_NAME was
getting very close to that and future commits might mimic the long name
and happen to
Hi Eelco,
Only 3 lines Changed so nothing much just a build failure was addressed
Regards
Amber
> -Original Message-
> From: Eelco Chaudron
> Sent: Tuesday, July 6, 2021 8:04 PM
> To: Ferriter, Cian
> Cc: ovs-dev@openvswitch.org; f...@sysclose.org; i.maxim...@ovn.org; Van
> Haaren,
Cian,
Which patches change, so I know where to update my review? None of the commit
messages show v6 changes.
//Eelco
On 6 Jul 2021, at 15:11, Cian Ferriter wrote:
> v6 updates:
> - Fix non-ssl build
> v5 updates:
> - reabse onto latest DPIF v14
> - use Enum for mfex impls
> - add pmd core id
See comments inline...
On 2 Jul 2021, at 14:56, Cian Ferriter wrote:
> From: Kumar Amber
>
> This patch introduces the mfex function pointers which allows
> the user to switch between different miniflow extract implementations
> which are provided by the OVS based on optimized ISA CPU.
>
> The
> -Original Message-
> From: Eelco Chaudron
> Sent: Friday, July 2, 2021 8:10 AM
> To: Van Haaren, Harry
> Cc: Amber, Kumar ; d...@openvswitch.org;
> i.maxim...@ovn.org; Flavio Leitner ; Stokes, Ian
>
> Subject: Re: [ovs-dev] [v4 02/12] dpif-netdev: Add auto validation function
> for
>
On 7/2/21 8:39 PM, Han Zhou wrote:
> On Fri, Jul 2, 2021 at 2:08 AM Dumitru Ceara wrote:
>>
>> Time how long change/run handlers take and log this at debug level.
>> I-P engine debug logs are not so verbose so enabling them is quite
>> common when debugging scale/control plane latency related
Time how long change/run handlers take and log this at debug level.
I-P engine debug logs are not so verbose so enabling them is quite
common when debugging scale/control plane latency related issues.
One of the major missing pieces though was a log about how long each I-P
node change/run handler
On Tue, Jul 06, 2021 at 03:27:41PM +0200, Adrian Moreno wrote:
>
>
> On 7/6/21 2:50 PM, Flavio Leitner wrote:
> > On Tue, Jul 06, 2021 at 08:25:59AM +0200, Adrian Moreno wrote:
> >>
> >>
> >> On 7/5/21 4:15 PM, Flavio Leitner wrote:
> >>>
> >>> Hi,
> >>>
> >>> On Wed, Jun 30, 2021 at 05:43:54PM
> -Original Message-
> From: Ilya Maximets
> Sent: Thursday, July 1, 2021 11:32 AM
> To: Van Haaren, Harry ; Ilya Maximets
>
> Cc: Eli Britstein ; ovs dev ; Ivan
> Malov
> ; Majd Dibbiny ; Stokes, Ian
> ; Ferriter, Cian ; Ben Pfaff
> ; Balazs Nemeth ; Sriharsha Basavapatna
>
> Subject:
On 7/6/21 2:50 PM, Flavio Leitner wrote:
> On Tue, Jul 06, 2021 at 08:25:59AM +0200, Adrian Moreno wrote:
>>
>>
>> On 7/5/21 4:15 PM, Flavio Leitner wrote:
>>>
>>> Hi,
>>>
>>> On Wed, Jun 30, 2021 at 05:43:54PM +0200, Adrian Moreno wrote:
The match keyword "igmp" is not supported in
Bleep bloop. Greetings Cian Ferriter, 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.
git-am:
error: sha1 information is lacking or useless (lib/dpif-netdev.c).
error: could not build fake
Bleep bloop. Greetings Paolo Valerio, 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.
git-am:
error: sha1 information is lacking or useless (lib/ovs-actions.xml).
error: could not build fake
From: Harry van Haaren
This commit adds 3 new traffic profile implementations to the
existing avx512 miniflow extract infrastructure. The profiles added are:
- Ether()/IP()/TCP()
- Ether()/Dot1Q()/IP()/UDP()
- Ether()/Dot1Q()/IP()/TCP()
The design of the avx512 code here is for scalability to
From: Harry van Haaren
This commit adds AVX512 implementations of miniflow extract.
By using the 64 bytes available in an AVX512 register, it is
possible to convert a packet to a miniflow data-structure in
a small quantity instructions.
The implementation here probes for Ether()/IP()/UDP()
From: Kumar Amber
Tests:
6: OVS-DPDK - MFEX Autovalidator
7: OVS-DPDK - MFEX Autovalidator Fuzzy
Added a new directory to store the PCAP file used
in the tests and a script to generate the fuzzy traffic
type pcap to be used in fuzzy unit test.
Signed-off-by: Kumar Amber
---
v5:
- fix
From: Harry van Haaren
This commit adds a new counter to be displayed to the user when
requesting datapath packet statistics. It counts the number of
packets that are parsed and a miniflow built up from it by the
optimized miniflow extract parsers.
The ovs-appctl command
From: Harry van Haaren
This commit enables OVS to at runtime check for more detailed
AVX512 capabilities, specifically Byte and Word (BW) extensions,
and Vector Bit Manipulation Instructions (VBMI).
These instructions will be used in the CPU ISA optimized
implementations of traffic profile
Hi Kevin ,
> -Original Message-
> From: Kevin Traynor
> Sent: Friday, July 2, 2021 4:55 AM
> To: d...@openvswitch.org
> Cc: david.march...@redhat.com; Pai G, Sunil ;
> cfont...@redhat.com; Kevin Traynor
> Subject: [PATCH v2 7/7] tests: Add new test for cross-numa pmd rxq
>
From: Kumar Amber
This commit introduces additional command line paramter
for mfex study function. If user provides additional packet out
it is used in study to compare minimum packets which must be processed
else a default value is choosen.
Also introduces a third paramter for choosing a
From: Kumar Amber
This commit adds a new command to allow the user to enable
autovalidatior by default at build time thus allowing for
runnig unit test by default.
$ ./configure --enable-mfex-default-autovalidator
Signed-off-by: Kumar Amber
Co-authored-by: Harry van Haaren
Signed-off-by:
From: Kumar Amber
This commit adds a section to the dpdk/bridge.rst netdev documentation,
detailing the added miniflow functionality. The newly added commands are
documented, and sample output is provided.
The use of auto-validator and special study function is also described
in detail as well
From: Kumar Amber
The study function runs all the available implementations
of miniflow_extract and makes a choice whose hitmask has
maximum hits and sets the mfex to that function.
Study can be run at runtime using the following command:
$ ovs-appctl dpif-netdev/miniflow-parser-set study
From: Kumar Amber
This patch introduced the auto-validation function which
allows users to compare the batch of packets obtained from
different miniflow implementations against the linear
miniflow extract and return a hitmask.
The autovaidator function can be triggered at runtime using the
From: Kumar Amber
This patch introduces the mfex function pointers which allows
the user to switch between different miniflow extract implementations
which are provided by the OVS based on optimized ISA CPU.
The user can query for the available minflow extract variants available
for that CPU by
v6 updates:
- Fix non-ssl build
v5 updates:
- reabse onto latest DPIF v14
- use Enum for mfex impls
- add pmd core id set paramter in set command
- get command modified to display the pmd thread for individual mfex functions
- resolved comments from Eelco, Ian, Flavio
- Use Atomic to get and set
> -Original Message-
> From: Ilya Maximets
> Sent: Friday, July 2, 2021 2:34 PM
> To: Van Haaren, Harry ; Eli Britstein
> ; d...@openvswitch.org; Ilya Maximets
> Cc: Ivan Malov ; Majd Dibbiny
> Subject: Re: [ovs-dev] [PATCH V7 13/13] netdev-dpdk-offload: Add vxlan pattern
> matching
Signed-off-by: Paolo Valerio
Acked-by: Aaron Conole
---
lib/dpif-netdev.c | 12 +++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 3835e1c9c..1f30fd686 100644
--- a/lib/dpif-netdev.c
+++ b/lib/dpif-netdev.c
@@ -8312,6
this patch introduces for the userspace datapath the handling
of rules like the following:
ct(commit,nat(src=0.0.0.0),...)
Kernel datapath already handle this case that is particularly
handy in scenarios like the following:
Given A: 10.1.1.1, B: 192.168.2.100, C: 10.1.1.2
A opens a connection
this macro is handy when it comes paste two tokens when one or both
are macros.
Rename CURSOR_JOIN() to OVS_JOIN() and move it to util.h so that it can
be reused.
Signed-off-by: Paolo Valerio
Acked-by: Gaetan Rivet
Acked-by: Aaron Conole
---
lib/cmap.h |5 +
lib/util.h |7 +++
when a packet gets dnatted and then recirculated, it could be possible
that it matches another rule that performs another nat action.
The kernel datapath handles this situation turning to a no-op the
second nat action, so natting only once the packet. In the userspace
datapath instead, when the
1/4 is a prereq for the series because it fixes an issue that prevents
OVN to use all-zero snat due to the way it builds the pipeline.
The patch 2/4 has been introduced in v6 as a prereq of 3/4.
All the versions up to v6 were about patch 3/4.
{2,3,4}/4 are untouched, so the acks are kept.
v8:
On Tue, Jul 06, 2021 at 08:25:59AM +0200, Adrian Moreno wrote:
>
>
> On 7/5/21 4:15 PM, Flavio Leitner wrote:
> >
> > Hi,
> >
> > On Wed, Jun 30, 2021 at 05:43:54PM +0200, Adrian Moreno wrote:
> >> The match keyword "igmp" is not supported in ofp-parse, which means
> >> that flow dumps cannot
Hi Mark,
David had some comments about the NEWS file, and I found an issue
on Windows below.
On Tue, Jul 06, 2021 at 05:31:11AM -0400, Mark Gray wrote:
> The Open vSwitch kernel module uses the upcall mechanism to send
> packets from kernel space to user space when it misses in the kernel
>
Hi Kevin ,
> -Original Message-
> From: Kevin Traynor
> Sent: Friday, July 2, 2021 4:55 AM
> To: d...@openvswitch.org
> Cc: david.march...@redhat.com; Pai G, Sunil ;
> cfont...@redhat.com; Kevin Traynor
> Subject: [PATCH v2 6/7] dpif-netdev: Allow pin rxq and non-isolate PMD.
>
On 7/6/21 2:06 PM, Eli Britstein wrote:
>
> On 7/6/2021 1:37 PM, Andrew Rybchenko wrote:
>> External email: Use caution opening links or attachments
>>
>>
>> Port ID should be obtained from physical device used to
>> create/destroy flow rules.
>>
>> Fixes: d8ad173fb9c1 ("netdev-offload-dpdk: Log
On 7/6/2021 1:37 PM, Andrew Rybchenko wrote:
External email: Use caution opening links or attachments
Port ID should be obtained from physical device used to
create/destroy flow rules.
Fixes: d8ad173fb9c1 ("netdev-offload-dpdk: Log testpmd format for flow
create/destroy.")
The one above
> This allows creating the match strings for each LB VIP exactly once,
> instead of once per datapath as it was before this change, reducing CPU
> usage in the ovn-northd event processing loop.
>
> On a scaled ovn-kubernetes-like deployment for 120 nodes, with 120
> gateway logical routers and
Port ID should be obtained from physical device used to
create/destroy flow rules.
Fixes: d8ad173fb9c1 ("netdev-offload-dpdk: Log testpmd format for flow
create/destroy.")
Signed-off-by: Andrew Rybchenko
---
v2:
- fix E-mail to be in lower case
lib/netdev-offload-dpdk.c | 4 ++--
1 file
Port ID should be obtained from physical device used to
create/destroy flow rules.
Fixes: d8ad173fb9c1 ("netdev-offload-dpdk: Log testpmd format for flow
create/destroy.")
Signed-off-by: Andrew Rybchenko
---
lib/netdev-offload-dpdk.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
This allows creating the match strings for each LB VIP exactly once,
instead of once per datapath as it was before this change, reducing CPU
usage in the ovn-northd event processing loop.
On a scaled ovn-kubernetes-like deployment for 120 nodes, with 120
gateway logical routers and 16K load
On 06/07/2021 10:31, Mark Gray wrote:
> This series proposes a new method of distributing upcalls
> to user space threads attempting to resolve a number of
> issues with the current method.
>
> v2 - Rebase
> Address Flavio's comments
> v3 - Add man page to automake
> v4 - Rebase and address
The Open vSwitch kernel module uses the upcall mechanism to send
packets from kernel space to user space when it misses in the kernel
space flow table. The upcall sends packets via a Netlink socket.
Currently, a Netlink socket is created for every vport. In this way,
there is a 1:1 mapping between
Fixes: 1579cf677fcb ("dpif-linux: Implement the API functions to allow multiple
handler threads read upcall.")
Signed-off-by: Mark Gray
Acked-by: Flavio Leitner
---
Notes:
v1 - Reworked based on Flavio's comments:
* Added "Fixes" tag
lib/dpif-netlink.c | 4 ++--
1 file changed,
'n_handlers' and 'n_revalidators' are declared as type 'size_t'.
However, dpif_handlers_set() requires parameter 'n_handlers' as
type 'uint32_t'. This patch fixes this type mismatch.
Signed-off-by: Mark Gray
Acked-by: Flavio Leitner
---
Notes:
v1 - Reworked based on Flavio's comments:
This series proposes a new method of distributing upcalls
to user space threads attempting to resolve a number of
issues with the current method.
v2 - Rebase
Address Flavio's comments
v3 - Add man page to automake
v4 - Rebase and address Flavio's comments
Mark Gray (3):
ofproto: change
Small nits on the NEWS update.
On Mon, Jul 5, 2021 at 3:39 PM Mark Gray wrote:
> diff --git a/NEWS b/NEWS
> index a2a2dcf95d7d..80b13e358685 100644
> --- a/NEWS
> +++ b/NEWS
> @@ -29,7 +29,12 @@ Post-v2.15.0
> - ovsdb-tool:
> * New option '--election-timer' to the 'create-cluster'
On Mon, Jul 5, 2021 at 6:57 PM Numan Siddique wrote:
>
> On Mon, Jul 5, 2021 at 12:12 PM Frode Nordahl
> wrote:
> >
> > On Wed, Jun 30, 2021 at 12:32 AM Numan Siddique wrote:
> > >
> > > On Thu, Jun 10, 2021 at 10:13 AM Frode Nordahl
> > > wrote:
> > > >
> > > > On Thu, Jun 10, 2021 at 1:46 PM
On 7/5/21 4:15 PM, Flavio Leitner wrote:
>
> Hi,
>
> On Wed, Jun 30, 2021 at 05:43:54PM +0200, Adrian Moreno wrote:
>> The match keyword "igmp" is not supported in ofp-parse, which means
>> that flow dumps cannot be restored. This patch prints the igmp match
>> in the accepted format
62 matches
Mail list logo