On 2/7/22 18:09, Eli Britstein via dev wrote:
> A flow may be modified after its initial offload failed. In this case,
> according to [1], the modification is handled as a flow add.
> For a vport flow "add", the orig_in_port should be provided.
> Keep that field in the flow struct, so it can be
On 3/22/22 14:35, Ilya Maximets wrote:
> OVS_WAIT_UNTIL() macro has only 2 arguments and doesn't check the
> output of the command, but bonding and route tests are trying to use
> it as if it was AT_CHECK macro. That makes checks in those tests
> mostly useless, since they are not actually
Hi Dumitru,
Thank you for the review! I'm going to send a v2 shortly with some changes.
> I'm not sure what the benefit is to have 'start' as 'const unsigned char
> *'; it could easily be 'const char *' and we wouldn't have to cast later
> on. Also, a comment might be useful, describing that
Bleep bloop. Greetings Han Zhou, 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: Line is 81 characters long (recommended limit is 79)
#708 FILE:
Bleep bloop. Greetings Han Zhou, 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: Line is 144 characters long (recommended limit is 79)
#161 FILE: ovn-sb.xml:1992:
Bleep bloop. Greetings Han Zhou, 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: Line is 139 characters long (recommended limit is 79)
#24 FILE: ovn-sb.xml:1958:
On Mon, Mar 21, 2022 at 9:50 AM Numan Siddique wrote:
>
> On Mon, Mar 21, 2022 at 12:26 PM Numan Siddique wrote:
> >
> > On Mon, Mar 21, 2022 at 12:21 PM Numan Siddique wrote:
> > >
> > > On Sun, Mar 13, 2022 at 8:01 PM Han Zhou wrote:
> > > >
> > > > Add a new action ct_lb_mark, which is the
Add two new actions push & pop for stack operations.
Signed-off-by: Han Zhou
---
include/ovn/actions.h | 8 +
lib/actions.c | 73 +++
ovn-sb.xml| 15 +
tests/ovn.at | 11 +++
utilities/ovn-trace.c | 64
1. The ovn-northd.8.xml didn't cover the flow in the lr_in_arp_resolve
stage.
2. The original test didn't really test the ecmp-symmetric-reply because
the option takes effect only for gateway routers. This patch set the
chassis to make the router a gateway router. Also, add the check
On Mon, Mar 21, 2022 at 9:49 AM Numan Siddique wrote:
>
> On Sun, Mar 13, 2022 at 7:58 PM Han Zhou wrote:
> >
> > Some NICs support HW offloading for datapath flows, but masked access to
> > the 128-bit ct_label field may prevent a flow being offloaded due to HW
> > limitations. OVN's use of
Add a new action ct_lb_mark, which is the same as ct_lb except that it
internally uses ct_mark to store the NAT flag, while ct_lb uses ct_label
for the same purpose. This will be used later to move the masked access
of ct_label to ct_mark while keeping the backward compatibility.
Signed-off-by:
The documentation of the ct_lb action wasn't up to date. Add the
description for backends and hash_fields.
Fixes: 5af304e7478a ("Support selection fields in load balancer.")
Signed-off-by: Han Zhou
---
ovn-sb.xml | 29 ++---
1 file changed, 18 insertions(+), 11
Some NICs support HW offloading for datapath flows, but masked access to
the 128-bit ct_label field may prevent a flow being offloaded due to HW
limitations. OVN's use of ct_label currently includes:
- ct_label.blocked (1 bit)
- ct_label.natted (1 bit)
- ct_label.ecmp_reply_port (16 bits)
-
Acked-by: Mark Michelson
On 3/22/22 13:37, Vladislav Odintsov wrote:
If rpm is built with libcapng (default), /etc/logrotate.d/ovn
config file user is SEDed in postinstall section with 'ovn ovn'.
If one run logrotate, next error occurs:
```
[root@host ~]# logrotate /etc/logrotate.d/ovn
error:
On Thu, Mar 10, 2022 at 2:42 PM wrote:
> From: Numan Siddique
>
> Presently, if the inport or outport is a peer port (of router port),
> then we skip sending the packet to conntrack by not setting the
> reg0[0]/reg0[1] bits. But the packet still goes through the
> stages - ls_in_acl_hint,
On 3/22/22 11:06, Numan Siddique wrote:
On Mon, Mar 14, 2022 at 4:26 AM Xavier Simonart wrote:
This will more or less double the number of test cases.
It is possible to select a reduce set of test cases using -k "keywords".
Keyword such as
dp-groups=yes
dp-groups=no
Hi Ilya,
Thanks for highlighting the Patch.
Replies Inline.
> -Original Message-
> From: Ilya Maximets
> Sent: Monday, March 21, 2022 10:26 PM
> To: d...@openvswitch.org; Stokes, Ian ; Van Haaren,
> Harry ; Ferriter, Cian
> ; Amber, Kumar
> Cc: i.maxim...@ovn.org; Toms Atteka
>
In glibc 2.32 sys_siglist is no longer exported. The MT-safe function
sigdescr_np() is now available for the same purpose.
Signed-off-by: Mike Pattrick
---
configure.ac | 1 +
lib/signals.c | 5 +
2 files changed, 6 insertions(+)
diff --git a/configure.ac b/configure.ac
index
If rpm is built with libcapng (default), /etc/logrotate.d/ovn
config file user is SEDed in postinstall section with 'ovn ovn'.
If one run logrotate, next error occurs:
```
[root@host ~]# logrotate /etc/logrotate.d/ovn
error: /etc/logrotate.d/ovn:9 unknown user 'ovn'
error: found error in
Hi Amber,
I am seeing a failure in one of the unit tests.
Test:
7: OVS-DPDK - MFEX Autovalidator Fuzzy FAILED (system-dpdk.at:289)
More details about the failure:
2022-03-16T16:07:05.234Z|2|dpif_netdev_extract(pmd-c21/id:101)|ERR|Autovalidation
for avx512_dot1q_ipv6_tcp failed
Hi Amber,
As part of reviewing this patch series, I wanted to first run/test the code to
familiarize myself with it.
As part of this, I ran in to some problems where the IPv6 packets I use aren't
parsed by the AVX512 implementations unless they are 106B or larger.
I had a look at why this is
On 3/22/22 17:26, David Marchand wrote:
> On Tue, Mar 22, 2022 at 2:36 PM Ilya Maximets wrote:
>>
>> OVS_WAIT_UNTIL() macro has only 2 arguments and doesn't check the
>> output of the command, but bonding and route tests are trying to use
>> it as if it was AT_CHECK macro. That makes checks in
Hi,
I will be talking about the refactoring and moving to new architecture for
P4-OVS in a public meeting for IPDK tomorrow. Please find the link below.
On Tue, Mar 22, 2022 at 3:41 AM Harold Huang wrote:
>
> On Fri, Mar 18, 2022 at 11:33 PM David Marchand
> wrote:
> >
> > DPDK based dp-packets points to data buffers that can't be expanded
> > dynamically.
> > Their layout is as follows:
> > - a 128 bytes headroom chosen at DPDK build time
On Mon, Mar 21, 2022 at 6:54 AM Peng He wrote:
>
> Hi,
> do you have future plan to support mbuf chaining for dp-packet?
> as I see you've actually relax some check on dp-packet in this
> patch.
To me, mbuf chaining means chaining packets, like having a list of packets.
So I guess you mean
Hi everyone,
Last month, commit d7514abe1 (acl-log: Log the direction (logical
pipeline) of the matching ACL.) was added to the main branch of OVN.
This commit was made because of the issue report at
https://bugzilla.redhat.com/show_bug.cgi?id=1992641 .
At the time the change was added, it
On Tue, Mar 22, 2022 at 2:36 PM Ilya Maximets wrote:
>
> OVS_WAIT_UNTIL() macro has only 2 arguments and doesn't check the
> output of the command, but bonding and route tests are trying to use
> it as if it was AT_CHECK macro. That makes checks in those tests
> mostly useless, since they are
Ilya Maximets writes:
> OVS_WAIT_UNTIL() macro has only 2 arguments and doesn't check the
> output of the command, but bonding and route tests are trying to use
> it as if it was AT_CHECK macro. That makes checks in those tests
> mostly useless, since they are not actually checking anything
s/alb.ut/alb.at/ in title.
On Fri, Mar 11, 2022 at 6:06 PM Kevin Traynor wrote:
>
> PMD auto load balance currently only operates when the polling pmd core
> will not change numa after reassignment. Add a unit test for this.
This could be squashed with the fix from patch 4.
>
> Signed-off-by:
On Fri, Mar 11, 2022 at 6:06 PM Kevin Traynor wrote:
>
> There are currently some checks for cross-numa polling cases to
> ensure that they won't effect the accuracy of the PMD ALB.
>
> If an rxq is pinned to a pmd core by the user it will not be
I prefer "PMD thread" for consistency (both in
On Fri, Mar 11, 2022 at 6:06 PM Kevin Traynor wrote:
>
> Ensure that if there are no local numa pmd cores
> available that pmd cores from all other non-local
> numas will be used.
This could be squashed with patch 2.
>
> Signed-off-by: Kevin Traynor
> ---
> tests/pmd.at | 62
On Fri, Mar 11, 2022 at 6:06 PM Kevin Traynor wrote:
>
Nit: the title is a bit scary, while this fix only matters when using
cores from (strictly) more than 2 numas in OVS.
To help users understand the impact of the issue/fix, I would add a
comment about the "strictly more than 2 numas" setup at
On Fri, Mar 11, 2022 at 6:06 PM Kevin Traynor wrote:
>
> Rename pmd_reblance_dry_run_needed() to
> pmd_rebalance_dry_run_needed().
>
> Fixes: a83a406096e9 ("dpif-netdev: Sync PMD ALB state with user commands.")
Nit: this nit applies to other patches in this series.
An empty line between Fixes:
I would like to request the implementation of this patch, with this patch I
will be able to implement my use case on OpenStack.
Thanks
Gui
From: Christophe Fontaine
Date: Thu, Mar 17, 2022 at 4:54 PM
Subject: [PATCH] ofproto/bond: Add knob "all_slaves_active"
To:
Cc: Christophe Fontaine
On Mon, Mar 14, 2022 at 4:26 AM Xavier Simonart wrote:
>
> This will more or less double the number of test cases.
> It is possible to select a reduce set of test cases using -k "keywords".
> Keyword such as
> dp-groups=yes
> dp-groups=no
> parallelization=yes
>
> On 22 Mar 2022, at 14:31, Mark Michelson wrote:
>
> No, that was not my intention. Basically, the following forms would never
> return an error:
>
> ovn-nbctl lr-nat-del my_router dnat (type only)
> ovn-nbctl lr-nat-del my_router dnat 172.16.0.1 (type and IP)
> ovn-nbctl lr-nat-del
On 3/22/22 09:46, Abhiram Sangana wrote:
On 21 Mar 2022, at 19:49, Mark Michelson wrote:
I'm a bit surprised that with this change, the hierarchy for specificity is TYPE <
IP < GATEWAY_PORT . This makes it sound as though the primary use case would be to
use the same IP for multiple NAT
As strongswan moved to the modern vici-based interface,this patch
modifies ovs-monitor-ipsec to use strongswan's vici-based
configuration instead of the legacy stroke-based configuration.
Reviewed-by: Raed Salem
Signed-off-by: Emeel Hakim
---
ipsec/ovs-monitor-ipsec.in | 466
patch ovs-monitor-ipsec: Migration from ipsec.conf to swanctl.conf
introduces strongswan-swanctl as an openvswitch-ipsec dependency,
this patch adds it to the openvswitch-ipsec dependencies.
Signed-off-by: Emeel Hakim
---
debian/control | 1 +
1 file changed, 1 insertion(+)
diff --git
As strongswan moved to the modern vici-based interface, this series migrates to
using swanctl.conf instead of ipsec.conf for configuring ipsec when
ovs-monitor-ipsec with strongswan as an IKE daemon, hence this series
introduces strongswan-swanctl as a new dependency for openvswitch-ipsec, this
> On 21 Mar 2022, at 19:49, Mark Michelson wrote:
>
> I'm a bit surprised that with this change, the hierarchy for specificity is
> TYPE < IP < GATEWAY_PORT . This makes it sound as though the primary use case
> would be to use the same IP for multiple NAT rules across different gateway
>
On 3/22/22 13:35, Ilya Maximets wrote:
> On 3/21/22 17:58, Ilya Maximets wrote:
>> On 3/21/22 15:45, David Marchand wrote:
>>> On Mon, Mar 21, 2022 at 1:39 PM Ilya Maximets wrote:
OVS_WAIT_UNTIL() macro has only 2 arguments and doesn't check
the output of the command, but bonding
Bleep bloop. Greetings Emeel Hakim, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error, please
Bleep bloop. Greetings Emeel Hakim, 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 failed
OVS_WAIT_UNTIL() macro has only 2 arguments and doesn't check the
output of the command, but bonding and route tests are trying to use
it as if it was AT_CHECK macro. That makes checks in those tests
mostly useless, since they are not actually checking anything except
for command returning zero.
As strongswan moved to the modern vici-based interface,this patch
modifies ovs-monitor-ipsec to use strongswan's vici-based
configuration instead of the legacy stroke-based configuration.
Reviewed-by: Raed Salem
Signed-off-by: Emeel Hakim
---
ipsec/ovs-monitor-ipsec.in | 459
patch ovs-monitor-ipsec: Migration from ipsec.conf to swanctl.conf
introduces strongswan-swanctl as an openvswitch-ipsec dependency,
this patch adds it to the openvswitch-ipsec dependencies.
Signed-off-by: Emeel Hakim
---
debian/control | 1 +
1 file changed, 1 insertion(+)
diff --git
As strongswan moved to the modern vici-based interface, this series
migrates to using swanctl.conf instead of ipsec.conf for configuring ipsec
when ovs-monitor-ipsec with strongswan as an IKE daemon, hence this series
introduces strongswan-swanctl as a new dependency for openvswitch-ipsec,
this
On 3/21/22 17:58, Ilya Maximets wrote:
> On 3/21/22 15:45, David Marchand wrote:
>> On Mon, Mar 21, 2022 at 1:39 PM Ilya Maximets wrote:
>>>
>>> OVS_WAIT_UNTIL() macro has only 2 arguments and doesn't check
>>> the output of the command, but bonding tests are trying to use
>>> it as if it was
If netdev goes away(i.e. qemu with a vnet shutdown), kernel delete all tc
filters, those tcf_id related to the netdev will be left in ufid_to_tc
hashmap. When qemu restart with a same vnet but different ifindex assigned,
a dup ufid may add. Especially after hashmap_expand, the old entry will
Since this patch moved to using strongswan’s vici-based configuration,
strongswan-swanctl becomes a dependency to openvswitch-ipsec package hence I
will be resending this patch as a series of two patches; this patch in addition
to a dependency update patch.
From: Mohammad Heib
Sent: Thursday,
The parse_key_and_mask_to_match() is a function to translate
a netlink formatted key/mask to match structure. And should
not consider any configuration setting when translating.
In addition we also enforce the encap_eth_type[0] mask as
it's required for the VLAN match.
Signed-off-by: Eelco
On 21 Mar 2022, at 10:07, Ilya Maximets wrote:
> On 2/17/22 13:27, Ilya Maximets wrote:
>> On 1/31/22 11:34, Eelco Chaudron wrote:
>>> The parse_key_and_mask_to_match() is a function to translate
>>> a netlink formatted key/mask to match structure. And should
>>> not consider any configuration
On Fri, Mar 18, 2022 at 1:49 PM Tonghao Zhang wrote:
>
> On Wed, Mar 9, 2022 at 10:20 PM Wan Junjie wrote:
> >
> > When several pmd threads race for the meter lock, the time 'now' each pmd
> > runs will
> > have no guarantee in the order they really happen.
> >
> > For example, pmd A runs meter
> Hi, Sunil.
>
> Mail-list strips attachments. So, if you want to share the presentation,
> upload it somewhere accessible and provide a link instead.
Hi ,
Thanks for the suggestion, Ilya.
Here is a publicly accessible link to the presentation:
55 matches
Mail list logo