Re: [PATCH printk v1 1/1] printk: nbcon: Allow unsafe write_atomic() for panic

[John Ogness <john.ogness@linutronix.de>]

On 2025-09-17, Petr Mladek <pmladek@suse.com> wrote:
>> After weighing the pros/cons I think that a global variable makes the
>> most sense. It will simplify internal APIs and provide all
>> console_is_usable() users a correct value. And the end result is no
>> different than what we do now.
>> 
>> We could also keep its setting inside nbcon_atomic_flush_unsafe() so
>> that the variable remains a printk-internal variable.
>
> Sounds good to me.

Right now things are a bit of a mess with required changes sitting in
printk and mm trees. Since this won't be going in to the upcoming merge
window, I will wait with v2 until you (Petr) can officially rebase the
printk tree to include the recent panic_*cpu*() changes. That will also
make it easier to coordinate the upcoming console_is_usable() changes as
well.

The functionality for v2 is the same as the v1, so the network folks can
continue working on the nbcon netconsole implementation.

@Breno: Or were you planning on pushing the nbcon netconsole for the 6.18
merge window next week? (I would guess no.)

John