Re: [PATCH 0/2] sched: Introduce and use deferred WARNs in sched

[Sebastian Andrzej Siewior]

On 2026-06-24 11:33:29 [+0200], Peter Zijlstra wrote:
> So I really don't see why we should do this. This has been a 'problem'
> forever, and printk() is actually being fixed.

I just prepared v2 and apparently missed this. So let me postpone it…

A user triggered a warning under the rq lock so it is not a theoretical
problem which happens during development. printk is fixed it is just the
console driver that are lagging behind. The 8250 is probably the role
model that everyone is waiting for and once it becomes a NBCON it is no
longer affected by the "deferred" requirement here.
The problem was reproduced on netconsole which is a NBCON, too.

So we just ignore the problem because it is well known? I was hoping for
some duct tape until we have NBCON and threaded-legacy consoles.

Vlad: Were you doing something special to trigger the warning in
question or did it randomly trigger?

Sebastian