[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <31b2e5c4-72cf-c9b8-2a0e-2d5800c59aac@arista.com>
Date: Tue, 28 May 2019 16:58:24 +0100
From: Dmitry Safonov <dima@...sta.com>
To: Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>,
Petr Mladek <pmladek@...e.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
linux-kernel@...r.kernel.org,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
Jiri Slaby <jslaby@...e.com>,
Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
Steven Rostedt <rostedt@...dmis.org>
Subject: Re: [RFC] printk/sysrq: Don't play with console_loglevel
Hi guys,
I see that the thread is ongoing and you understand printk code much
better than me or probably anybody :)
So, feel free to reuse it. Or I can send v1 with split sysrq/printk
parts if you think it's worth being shaped further.
I think worth to mention three "features" that you might had a chance to
miss from the current code:
1. op_p->handler(key) is not printed under console_loglevel hackery.
So, under RFC I preserved the behavior. Probably, you don't miss it
and just looking into ways to change it, but I thought worth
mentioning.
2. I've found it surprising how pr_info() interacts with pr_cont():
Basically, pr_cont() without KERN_<LEVEL> prefix will print the
resulting line with default_message_loglevel AFAIU from cont.level.
Which might be higher than warning.
I might miss a part that corrects cont.level to the first
message's level?
3. CONSOLE_LOGLEVEL_DEFAULT is config-based, so having in mind that it
can be changed and (2) - it gives me hard time to understand when
the sysrq message will be printed and when will be suppressed.
Thanks,
Dmitry
Powered by blists - more mailing lists