Re: [PATCH v2 0/3] net: dsa: SERDES support for mv88e632x family

[Sverdlin, Alexander]

Hello Andrew,

On Thu, 2023-07-13 at 21:33 +0200, Andrew Lunn wrote:
> On Thu, Jul 13, 2023 at 04:29:12PM +0000, Sverdlin, Alexander wrote:
> > Hi Andrew,
> > 
> > On Thu, 2023-07-13 at 17:34 +0200, Andrew Lunn wrote:
> > > > we did understand Andrew's request indeed, however we were not able to
> > > > backport your series quickly to the version we are able to test on the
> > > > HW.
> > > 
> > > But your own patchset has been tested on net-next by yourself? So it
> > > should be trivial to use "b4 am <threadid>" to get Russells patchset,
> > > apply them on top of net-next, and then add your patches as well.
> > 
> > unfortunately it has been only built on net-next and tested on a 5.15.x with some
> > limited backports so that Russell series would require even more backporting.
> 
> You should indicate that the patch is only built tested when
> submitting it.

yes, it's not tested on the HW with net-net.
It's tested with older kernel with DSA backported to 6.1+ kernel.

> I assume you are using a vendor kernel? What is missing in mainline to
> stop networking working?

If we talk networking, my patch "net: fec: Defer probe if other FEC has deferred MDIO"
comes to my mind, this is quite important, but didn't gain much interest.

But network is usually not the critical part, it's usually other SoC parts and it's
not always easy to plug a Gigabit Switch out and plug it into a HW running fresh kernel.

We are improving on this, but this is a balancing act between being creative on
testing and not upstreaming at all.

-- 
Alexander Sverdlin
Siemens AG
www.siemens.com