Re: [PATCH v3 2/3] printk: Fix preferred console selection with multiple matches

[Petr Mladek]

On Thu 2020-02-06 15:02:25, Benjamin Herrenschmidt wrote:
> In the following circumstances, the rule of selecting the console
> corresponding to the last "console=" entry on the command line as
> the preferred console (CON_CONSDEV, ie, /dev/console) fails. This
> is a specific example, but it could happen with different consoles
> that have a similar name aliasing mechanism.
> 
> This tentative fix register_console() to scan first for consoles
> specified on the command line, and only if none is found, to then
> scan for consoles specified by the architecture.
> 
> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
> index 17602d7b7ffc..5cf47a7b880c 100644
> --- a/kernel/printk/printk.c
> +++ b/kernel/printk/printk.c
> @@ -2674,8 +2679,13 @@ static int try_enable_new_console(struct console *newcon)
>  	/*
>           * Some consoles, such as pstore and netconsole, can be enabled even
>           * without matching.
> +	 *
> +	 * Note: We only do this test on the !user_specified pass so that such
> +	 * a statically enabled console that isn't user specified gets a chance
> +	 * to have its match() or setup() function called on our second pass
> +	 * through this function.

I had some troubles to part the comment. I wonder if the following is
more clear:

	* Accept pre-enabled consoles only when match() and setup()
	* was called.

And I would do the same check as in the for cycle:

	if (newcon->flags & CON_ENABLED && c->user_specified ==	user_specified)
		return 0;

With the above change:

Reviewed-by: Petr Mladek <pmladek@xxxxxxxx>

I could do the change when pushing if you agree and v4 is not needed
for other reasons.

Best Regards,
Petr

PS: JFYI, I am going to look at the 3rd patch tomorrow. I have to go now.