Date: Mon, 25 Apr 2016 15:35:26 -0400
From: Steven Rostedt <rostedt@goodmis.org>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>, LKML <linux-kernel@vger.kernel.org>,
        Ingo Molnar <mingo@kernel.org>, Borislav Petkov <bp@alien8.de>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Uwe =?UTF-8?B?S2xlaW5lLUvDtm5p?= =?UTF-8?B?Zw==?= 
	<u.kleine-koenig@pengutronix.de>
Subject: Re: [PATCH v2] printk: Add kernel parameter to disable writes to
 /dev/kmsg
Message-ID: <20160425153526.284cb76a@gandalf.local.home>
In-Reply-To: <CA+55aFzKP4N0xSZGF4kkTcK7iOyAmOK+CR1B=DfD8=TsN0qnEA@mail.gmail.com>
References: <20160425145606.598329f2@gandalf.local.home>
	<20160425190651.GL3448@twins.programming.kicks-ass.net>
	<CA+55aFzKP4N0xSZGF4kkTcK7iOyAmOK+CR1B=DfD8=TsN0qnEA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1367
Lines: 32

On Mon, 25 Apr 2016 12:28:30 -0700
Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Mon, Apr 25, 2016 at 12:06 PM, Peter Zijlstra <peterz@infradead.org> wrote:
> >
> > Again, please default enable and use an easier name to toggle this.
> > Userspace flooding this with junk is really insane.  
> 
> I think it should be a tristate with "yes/no/ratelimit", and let's
> default to ratelimit.

Hmm, as this is currently decided on opening, it would be an
interesting task, and more complex. But I could look into it.

> 
> And I also suspect that we would be better off not returning an error
> (which could make user space decide to break, either intentionally or
> just because some people think that "error handling is important"
> means that you should abort on all errors you don't recognize), but
> just silently drop the write. IOW, the "no" would just be a rather
> extreme form of rate-limiting, while "yes" would just be the other
> extreme.
> 

Actually, systemd currently does the right thing when this errors on
opening. It finds another way to do logging. At least on my boxes that
I tested (Debian testing and Fedora 18). If it doesn't error out,
systemd may decided to continue to use this as its means for general
purpose logging, and now I lose out on the logging of userspace, as
everything that systemd does is silently dropped.

-- Steve