[ovs-dev] [PATCH v4 0/5] Backport upstream mpls fix

[Joe Stringer]

On 3 May 2017 at 08:50, Simon Horman <simon.horman at netronome.com> wrote:
> On Tue, May 02, 2017 at 10:38:17AM -0700, Joe Stringer wrote:
>> On 2 May 2017 at 05:03, Simon Horman <simon.horman at netronome.com> wrote:
>> > On Mon, May 01, 2017 at 10:24:30AM -0700, Yi-Hung Wei wrote:
>> >> This series backports the upstream mpls fixes and add a system traffic
>> >> test for mpls actions.
>> >
>> > Thanks Yi-Hung, this looks good to me.
>> >
>> > Acked-by: Simon Horman <simon.horman at netronome.com>
>> >
>> > Joe, were you considering applying this?
>> > I'm happy to do so myself if there are no objections.
>>
>> Thanks for reviewing these, Simon. I also took a look and they seem
>> fine to me. Please do apply them, it'll be good to get this resolved.
>
> Thanks, I have pushed these to master.
> Should the also be considered for branch-2.6, branch-2.5, etc...?

Does it cleanly apply? If it does, then we could just apply them. I
know that for instance patch 8bc50076948b ("datapath: Fix
ovs_flow_key_update()") is only relevant for master because the l3
tunneling changes weren't in any prior release.

Some of these fixes, for example handling the different pointer
layouts for L2/MPLS/inner, are probably only relevant for kernels that
are newer than the supported kernels for older releases. branch-2.7
might be an exception, could you take a look?

It's nice to be able to fix all the bugs back to the last LTS (in this
case 2.5), but in practice it seems like no-one has noticed these
breakages in any release back to 2.4 so I'm not sure it's worth the
effort. Perhaps the people who are using MPLS are not using the OVS
out-of-tree module provided with the OVS userspace distribution, in
which case they aren't affected. Or perhaps they're not using GSO, or
they don't load the MPLS GSO module.