Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   John Ogness <john.ogness@linutronix.de>
To:     Petr Mladek <pmladek@suse.com>
Cc:     Sergey Senozhatsky <senozhatsky@chromium.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        linux-kernel@vger.kernel.org, Helge Deller <deller@gmx.de>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Javier Martinez Canillas <javierm@redhat.com>,
        Thomas Zimmermann <tzimmermann@suse.de>,
        Juergen Gross <jgross@suse.com>,
        Boris Ostrovsky <boris.ostrovsky@oracle.com>,
        Tom Rix <trix@redhat.com>, linux-fbdev@vger.kernel.org,
        dri-devel@lists.freedesktop.org
Subject: Re: [PATCH printk v3 33/40] printk, xen: fbfront: create/use safe
 function for forcing preferred
In-Reply-To: <Y20aBwNWT19YDeib@alley>
References: <20221107141638.3790965-1-john.ogness@linutronix.de>
 <20221107141638.3790965-34-john.ogness@linutronix.de>
 <Y20aBwNWT19YDeib@alley>
Date:   Thu, 10 Nov 2022 17:09:12 +0106
Message-ID: <877d026blr.fsf@jogness.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

On 2022-11-10, Petr Mladek <pmladek@suse.com> wrote:
>> +void console_force_preferred_locked(struct console *con)
>> +{
>> +	struct console *cur_pref_con;
>> +
>> +	if (!console_is_registered_locked(con))
>> +		return;
>> +
>> +	cur_pref_con = console_first();
>> +
>> +	/* Already preferred? */
>> +	if (cur_pref_con == con)
>> +		return;
>> +
>> +	hlist_del_init_rcu(&con->node);
>
> We actually should re-initialize the node only after all existing
> console list walks are finished. Se we should use here:
>
> 	hlist_del_rcu(&con->node);

hlist_del_init_rcu() only re-initializes @pprev pointer. But maybe you
are concerned that there is a window where list_unhashed() becomes true?
I agree that it should be changed to hlist_del_rcu() because there
should not be a window where this console appears unregistered.

>> +	/* Only the new head can have CON_CONSDEV set. */
>> +	WRITE_ONCE(cur_pref_con->flags, cur_pref_con->flags & ~CON_CONSDEV);
>
> As mentioned in the reply for 7th patch, I would prefer to hide this
> WRITE_ONCE into a wrapper, e.g. console_set_flag(). It might also
> check that the console_list_lock is taken...

Agreed. For v4 it will become:

console_srcu_write_flags(cur_pref_con->flags & ~CON_CONSDEV);

John