Received-SPF: pass (google.com: domain of linux-kernel+bounces-214473-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Fri, 14 Jun 2024 09:40:38 +0200
From: Petr Mladek <pmladek@suse.com>
To: John Ogness <john.ogness@linutronix.de>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>,
	Steven Rostedt <rostedt@goodmis.org>,
	Thomas Gleixner <tglx@linutronix.de>, linux-kernel@vger.kernel.org,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Subject: Re: [PATCH printk v2 04/18] printk: nbcon: Introduce printing
 kthreads
Message-ID: <Zmvz4gC12ekBY9mz@pathway.suse.cz>
References: <20240603232453.33992-1-john.ogness@linutronix.de>
 <20240603232453.33992-5-john.ogness@linutronix.de>
 <ZmMIOWEAMgXz4j-U@pathway.suse.cz>
 <87ed95j8yh.fsf@jogness.linutronix.de>
 <ZmhkVAC_3FMohrEr@pathway.suse.cz>
 <87sexipmrk.fsf@jogness.linutronix.de>
 <87jzis50n6.fsf@jogness.linutronix.de>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <87jzis50n6.fsf@jogness.linutronix.de>

On Thu 2024-06-13 17:27:33, John Ogness wrote:
> On 2024-06-11, Petr Mladek <pmladek@suse.com> wrote:
> > Honestly, if the system is not able to start the kthread then
> > it is probably useless anyway. I would prefer if printk keeps working
> > so that people know what is going on ;-)
> 
> I have been looking into and thinking about this some more. I do not
> like the idea of the system doing some sort of fallback if some of the
> kthreads fail to start. Not only does it complicate the code, but it
> greatly increases the variants of how the system during runtime.

Fair enough.

> I (strongly) suggest the following:
> 
> - The kthread is created in nbcon_alloc(). If the kthread fails, then
>   nbcon_alloc() fails and the console will not register.
>
> - nbcon_kthread_should_wakeup() will return false if the console is not
>   registered or if @have_boot_console=1.

I like this.

I guess that this would require moving some pieces from nbcon_init
to nbcon_alloc(). It might make sense to move there everything except
for setting the initial seq number.

> Then there would be no need to ever check con->kthread. Instead we can
> rely on the global state of the system transitioning to relying on
> threading.
> 
> I think it is totally appropriate that register_console() fails if the
> kthread cannot be created, just as it already fails if the kmalloc() for
> the pbufs fails.
> 
> Any objections?

Not from my side ;-)

Best Regards,
Petr