Re: Old platforms: bring out your dead

On Mon, Jan 11, 2021 at 11:27 AM Alexander Sverdlin
[off-list ref] wrote:

On 11/01/2021 10:31, Arnd Bergmann wrote:

quoted

quoted

quoted

* axxia -- added in 2014, no notable changes after 2015

This one is still widely used there is a chance you are using it too when
traveling in Asia or USA :)

I know this one has a large installed base, and that Intel were at
least working on it for a while after upstream contributions stopped.

However, my impression was that the port was never completed
upstream before the acquisition, and that the new owners had no
interest in working with us. In particular, none of the later Axxia
SoCs (ppc32/axe3500, arm64/axm5600, arm64/axc6700) ever got
submitted for inclusion as far as I can tell.

The latest public source code I found is at https://github.com/axxia/,
but this is a heavily patched linux-4.9, see
https://github.com/axxia/axxia_yocto_linux_4.9/commits/fa03c456641

My interpretation of this was that whoever is using Axxia chips
is no longer interested in upgrading to newer kernels or using
anything remotely resembling the code we have uptream.

we indeed upgrade to the latest kernels but yes, we have to forward-port
LSI/Intel's code again and again.

And while AXM55xx is ugly hardware in itself where we need to patch a lot,
64-bit variants are working out of the box with vanilla kernel actually.

We by no means want this platform to be thrown away, so let us work a
road map out.

I'll assess what the situation is with AXM55xx, AXM56xx and AXC67
(which from my PoV
is no different from AXM56xx) and share my ideas.

Sure, sounds good. Great to head the 64-bit parts actually work.
It would be great to have a Kconfig.platforms entry for it and
maybe one .dts file either for a reference machine, or one of the
products you work with.

I'm of course not in the position to submit patches you found in their
github account, but maybe there is another solution.

Why not? As long as the code looks reasonable and has a GPLv2
compatible license, you should be able to take any small or large
subset of their work and send it our way.

       Arnd

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel