Re: WARNING at tty_buffer.c:428 process_one_work()

[Jiri Slaby <jslaby@suse.cz>]

On 03/05/2013 08:39 PM, David Miller wrote:
> From: Jiri Slaby <jslaby@suse.cz>
> Date: Tue, 05 Mar 2013 12:01:06 +0100
> 
>> I left that "if (port->start == NULL)" in sunhv in place because it
>> behaves completely differently. It checks port->start on all paths prior
>> dereferencing it. And it does not stop interrupts on ->shutdown.
> 
> But this code really does care if a TTY is attached, because it wants
> to make sure that SYSRQ handling occurs unconditionally, even if there
> is no TTY to queue the characters to.
> 
> This is critically important during bootup before the initial shell
> is spawned, if you want to do a SYSRQ register dump or reset out of
> a hung boot.
> 
> Whether that test is now ->state == NULL or whatever, the same logic
> still needs to exist in all of these places.

Hi, I must admit I don't understand. I now checked both of them and they
call uart_handle_sysrq_char unconditionally, or?

-- 
js
suse labs