[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <aTbTAfJtVeOgtHAr@pathway>
Date: Mon, 8 Dec 2025 14:30:41 +0100
From: Petr Mladek <pmladek@...e.com>
To: John Ogness <john.ogness@...utronix.de>
Cc: Marcos Paulo de Souza <mpdesouza@...e.com>,
Steven Rostedt <rostedt@...dmis.org>,
Sergey Senozhatsky <senozhatsky@...omium.org>,
linux-kernel@...r.kernel.org, marcos@...esouza.com
Subject: Re: [PATCH] printk: nbcon: Check for device_{lock,unlock} callbacks
On Mon 2025-12-08 10:21:09, John Ogness wrote:
> On 2025-12-05, Marcos Paulo de Souza <mpdesouza@...e.com> wrote:
> > These callbacks are necessary to synchronize ->write_thread callback
> > against other operations using the same device.
> >
> > Signed-off-by: Marcos Paulo de Souza <mpdesouza@...e.com>
> > ---
> > I found this issue while creating a custom kernel module that implements
> > the nbcon interfaces.
Great catch!
> > --- a/kernel/printk/nbcon.c
> > +++ b/kernel/printk/nbcon.c
> > @@ -1764,6 +1764,10 @@ bool nbcon_alloc(struct console *con)
> > if (WARN_ON(!con->write_thread))
> > return false;
> >
> > + /* The device_lock() and device_unlock() callbacks are mandatory. */
> > + if (WARN_ON(!con->device_lock || !con->device_unlock))
> > + return false;
> > +
>
> The code itself is fine and this is an important contribution. But I
> would prefer to combine it with the previous mandatory check. So it
> looks more like this:
>
> diff --git a/kernel/printk/nbcon.c b/kernel/printk/nbcon.c
> index 3fa403f9831f..f096282b0625 100644
> --- a/kernel/printk/nbcon.c
> +++ b/kernel/printk/nbcon.c
> @@ -1760,9 +1758,12 @@ bool nbcon_alloc(struct console *con)
> /* Synchronize the kthread start. */
> lockdep_assert_console_list_lock_held();
>
> - /* The write_thread() callback is mandatory. */
> - if (WARN_ON(!con->write_thread))
> + /* Check for mandatory nbcon callbacks. */
> + if (WARN_ON(!con->write_thread ||
> + !con->device_lock ||
> + !con->device_unlock)) {
> return false;
> + }
I agree that this looks better. ;-)
>
> rcuwait_init(&con->rcuwait);
> init_irq_work(&con->irq_work, nbcon_irq_work);
Best Regards,
Petr
Powered by blists - more mailing lists