On Mon, Dec 19, 2022 at 03:39:50PM +0100, Ilya Maximets wrote: > On 12/17/22 07:15, Nole Zhang wrote: > > > > > >> -----邮件原件----- > >> 发件人: David Marchand <david.march...@redhat.com> > >> 发送时间: 2022年12月17日 4:02 > >> 收件人: Simon Horman <simon.hor...@corigine.com> > >> 抄送: d...@openvswitch.org; Eli Britstein <el...@nvidia.com>; Chaoyong He > >> <chaoyong...@corigine.com>; oss-drivers <oss-driv...@corigine.com>; Ilya > >> Maximets <i.maxim...@ovn.org>; Nole Zhang <peng.zh...@corigine.com> > >> 主题: Re: [ovs-dev] [PATCH 6/6] ci: add the opts about > >> ALLOW_EXPERIMENTAL_API > >> > >> [You don't often get email from david.march...@redhat.com. Learn why this > >> is important at https://aka.ms/LearnAboutSenderIdentification ] > >> > >> On Fri, Dec 16, 2022 at 4:52 PM Simon Horman <simon.hor...@corigine.com> > >> wrote: > >>> > >>> From: Peng Zhang <peng.zh...@corigine.com> > >>> > >>> This commit adds support for OVS-DPDK with > >> -DALLOW_EXPERIMENTAL_API. > >>> > >>> Tunnel offloads and Meter offloads are experimental APIs in DPDK. To > >>> enable these features, compile need add -DALLOW_EXPERIMENTAL_API. So > >>> in workflow, we also need need the new test with > >>> -DALLOW_EXPERIMENTAL_API. > >>> > >>> Signed-off-by: Peng Zhang <peng.zh...@corigine.com> > >> > >> We have a similar patch in the dpdk-latest branch. > >> https://github.com/openvswitch/ovs/commit/a8f6be98801f0c43d52173843d > >> 649df2af5e1c0d > >> Is something wrong with it? > > > > The patch is good for me, I just didn't notice it，thanks for your notice. > > I think, the main thing is that this patch set needs to be posted > against dpdk-latest branch, i.e. has the '[PATCH dpdk-latest]' > subject prefix. Changes that are using experimental DPDK features > are supposed to be developed and can be accepted in that branch.
Thanks Ilya, got it. Will do so with v2. > We did an exception in the past and accepted experimental tunnel > offloading support because it required extensive changes in many > generic parts of OVS and it would be a burden trying to maintain it > separately. But the time showed that it wasn't a good decision. > I'm actually considering a possibility of removing that support > because current DPDK API for tunnel offloading is not usable in > most cases [1]. It requires changes, but not going anywhere AFAIK. > > [1] https://inbox.dpdk.org/dev/5248c2ca-f2a6-3fb0-38b8-7f659bfa4...@ovn.org/ > > Best regards, Ilya Maximets. _______________________________________________ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev
