Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 24 Mar 2023 13:06:59 -0400
From:   Steven Rostedt <rostedt@goodmis.org>
To:     Beau Belgrave <beaub@linux.microsoft.com>
Cc:     Vlastimil Babka <vbabka@suse.cz>,
        "Masami Hiramatsu (Google)" <mhiramat@kernel.org>,
        mathieu.desnoyers@efficios.com, dcook@linux.microsoft.com,
        alanau@linux.microsoft.com, brauner@kernel.org,
        akpm@linux-foundation.org, ebiederm@xmission.com,
        keescook@chromium.org, tglx@linutronix.de,
        linux-trace-devel@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-mm@kvack.org
Subject: Re: [PATCH v8 11/11] tracing/user_events: Limit global user_event
 count
Message-ID: <20230324130659.587ecfd2@gandalf.local.home>
In-Reply-To: <20230324164353.GA1790@kbox>
References: <20230221211143.574-1-beaub@linux.microsoft.com>
        <20230221211143.574-12-beaub@linux.microsoft.com>
        <20230324081824.b917c2944da217e5239e1223@kernel.org>
        <d6c83572-17e1-93d4-65a0-d480989e54fb@suse.cz>
        <20230324164353.GA1790@kbox>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Fri, 24 Mar 2023 09:43:53 -0700
Beau Belgrave <beaub@linux.microsoft.com> wrote:

> > It was actually merged in 5.8. So sysctl should be sufficient with that.
> > But maybe it's weird to start adding sysctls, when the rest of tracing
> > tunables is AFAIK under /sys/kernel/tracing/ ?
> >   
> 
> During the TraceFS meetings Steven runs I was asked to add a boot
> parameter and sysctl for user_events to limit the max.
> 
> To me, it seems when user_events moves toward namespace awareness
> sysctl might be easier to use from within a namespace to turn knobs.
> 
> Happy to change to whatever, but I want to see Steven and Masami agree
> on the approach before doing so.
> 
> Steven, do you agree with Masami to move to just sysctl?

We do have some tracing related sysctls already:

# cd /proc/sys/kernel
# ls *trace*
ftrace_dump_on_oops  oops_all_cpu_backtrace  traceoff_on_warning
ftrace_enabled       stack_tracer_enabled    tracepoint_printk

Although I would love to deprecated ftrace_enable as that now has a
control in tracefs, but it's not unprecedented to have tracing tunables as
sysctl.

And if we get cmdline boot parameters for free from sysctls then all the
better.

-- Steve