Re: [Linux-kernel] [PATCH] tty: serial: 8250: add CON_CONSDEV to flags

[Jiri Slaby]

On 10/18/2016, 06:23 PM, Ben Dooks wrote:
> On 18/10/16 17:15, Jiri Slaby wrote:
>> On 06/21/2016, 05:19 PM, Matthew Leach wrote:
>>> When using the 8250 as a boot console and the main console results in
>>> messages being printed twice.  The console framework will only
>>> unregister boot consoles if a new console is registered with the
>>> CON_CONSDEV flag set.
>>>
>>> Set this flag for the univ8250 console to prevent double-registration.
>>>
>>> Signed-off-by: Matthew Leach <matthew.leach@xxxxxxxxxxxxxxx>
>>> ---
>>>  drivers/tty/serial/8250/8250_core.c | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/tty/serial/8250/8250_core.c
>>> b/drivers/tty/serial/8250/8250_core.c
>>> index 2f4f5ee..7a7453d 100644
>>> --- a/drivers/tty/serial/8250/8250_core.c
>>> +++ b/drivers/tty/serial/8250/8250_core.c
>>> @@ -675,7 +675,7 @@ static struct console univ8250_console = {
>>>      .device        = uart_console_device,
>>>      .setup        = univ8250_console_setup,
>>>      .match        = univ8250_console_match,
>>> -    .flags        = CON_PRINTBUFFER | CON_ANYTIME,
>>> +    .flags        = CON_PRINTBUFFER | CON_ANYTIME | CON_CONSDEV,
>>
>> This patch breaks console= parameter handling. Passing:
>>   console=ttyS0 console=tty0
>> no longer means /dev/console is tty0 and it should be as is documented.
>>
>> I have not investigated further yet.
> 
> Ok, thanks.
> I'll need to find Matthew Leach's new address as he's moved jobs.

Oh sure:
#define CON_CONSDEV     (2) /* Last on the command line */

Why it should be set in struct console statically? The patch is
obviously incorrect. And I do not understand from the vague commit log
what the problem actually was.

So for the time being, Greg, please revert it (d03516df8375873) and we
can come with something better, when we understand the issue. Please CC
stable (4.8+).

thanks,
-- 
js
suse labs