Re: [linux-next:master 70/7526] undefined reference to `__trace_hardirqs_off'

[Arnd Bergmann <arnd@arndb.de>]

On Thu, Feb 24, 2022 at 4:40 AM Steven Rostedt <rostedt@goodmis.org> wrote:
> On Wed, 23 Feb 2022 19:12:45 -0800 Randy Dunlap <rdunlap@infradead.org> wrote:
> > On 2/23/22 18:33, Steven Rostedt wrote:
> > > On Wed, 23 Feb 2022 17:53:00 -0800
> > > Randy Dunlap <rdunlap@infradead.org> wrote:
> > >
> > >> On 2/23/22 17:44, Randy Dunlap wrote:
> > >>> Hi ktr, bouncing NDS32 maintainer, Andrew, Steven,
> > >>>
> > >>
> > >> [now adding Andrew and Steven!]
> > >
> > > This needs to go via the NDS32 maintainers. If it's being ignored, then
> > > we can consider the architecture as orphaned and start the process of
> > > removal. ;-)
> >
> > Yes, it appears to be about that time...
>
> Hi Arnd,
>
> Just putting this on your "chopping block" radar.

Yes, that makes sense. I think nds32 was always going to be a
short-lived architecture as Andestech was already in the process of
moving to RISC-V before it was ever submitted upstream. I did
hope that it would live a little longer, but it seems that everyone who
ever worked on it in mainline has left andestech by now.

I'm guessing Nick's current email address based on his Linkedin page,
maybe he can contact someone who is still there and able to
take over.

I think nds32 is still in a better shape than a lot of the older
architectures in mainline, and the CPU cores are still actively
marketed for licensing. OTOH it only gets worse if nobody cares
about the code, and users are probably already better off with
one of the older longterm supported kernels. If we remove
arch/nds32 before the next longterm release, that gives users
stable updates until late 2026 using linux-5.10.y, which is
probably sufficient.

        Arnd