[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 2 Feb 2024 10:26:59 +0100
From: Petr Mladek <pmladek@...e.com>
To: John Ogness <john.ogness@...utronix.de>
Cc: Sergey Senozhatsky <senozhatsky@...omium.org>,
Steven Rostedt <rostedt@...dmis.org>,
Thomas Gleixner <tglx@...utronix.de>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH printk v3 13/14] printk: Avoid non-panic CPUs writing to
ringbuffer
On Thu 2023-12-14 22:48:00, John Ogness wrote:
> Commit 13fb0f74d702 ("printk: Avoid livelock with heavy printk
> during panic") introduced a mechanism to silence non-panic CPUs
> if too many messages are being dropped. Aside from trying to
> workaround the livelock bugs of legacy consoles, it was also
> intended to avoid losing panic messages. However, if non-panic
> CPUs are writing to the ringbuffer, then reacting to dropped
> messages is too late.
>
> To avoid losing panic CPU messages, silence non-panic CPUs
> immediately on panic.
>
> Signed-off-by: John Ogness <john.ogness@...utronix.de>
I am slightly nervous about this change because it looks
too agresive ;-)
But it makes perfect sense. And it nicely complements the 10th
patch, see
https://lore.kernel.org/r/20231214214201.499426-11-john.ogness@linutronix.de
The 10th patch allows to skip messages in reserved state. It might
cause skipping random messages from other CPUs. So it really
looks better to skip/ignore them completely.
It would be nice to mention the relation to the 10th patch
in the commit message. Something like:
<proposal>
Another motivation is that non-finalized messages already might be
skipped in panic(). By other words, random messages from non-panic
CPUs might already get lost. It is better to ignore all to avoid
confusion.
</proposal>
With the updated commit message:
Reviewed-by: Petr Mladek <pmladek@...e.com>
Best Regards,
Petr
Powered by blists - more mailing lists