From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1757597Ab3GERbw (ORCPT <rfc822;w@1wt.eu>);
	Fri, 5 Jul 2013 13:31:52 -0400
Received: from mx1.redhat.com ([209.132.183.28]:21922 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1752029Ab3GERbv (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Fri, 5 Jul 2013 13:31:51 -0400
Date: Fri, 5 Jul 2013 19:26:55 +0200
From: Oleg Nesterov <oleg@redhat.com>
To: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Steven Rostedt <rostedt@goodmis.org>, linux-kernel@vger.kernel.org,
        "zhangwei(Jovi)" <jovi.zhangwei@huawei.com>,
        Jiri Olsa <jolsa@redhat.com>, Peter Zijlstra <a.p.zijlstra@chello.nl>,
        Arnaldo Carvalho de Melo <acme@ghostprotocols.net>,
        Srikar Dronamraju <srikar@linux.vnet.ibm.com>,
        Frederic Weisbecker <fweisbec@gmail.com>,
        Ingo Molnar <mingo@redhat.com>,
        Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [RFC][PATCH 3/4] tracing/kprobes: Fail to unregister if probe
	event files are open
Message-ID: <20130705172655.GA18576@redhat.com>
References: <20130704033347.807661713@goodmis.org> <20130704034038.819592356@goodmis.org> <51D56E74.3090404@hitachi.com> <20130704184813.GA2695@redhat.com> <51D6351A.2000206@hitachi.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <51D6351A.2000206@hitachi.com>
User-Agent: Mutt/1.5.18 (2008-05-17)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 07/05, Masami Hiramatsu wrote:
>
> (2013/07/05 3:48), Oleg Nesterov wrote:
> > On 07/04, Masami Hiramatsu wrote:
> >>
> >> Actually disable_kprobe() doesn't ensure to finish the current running
> >> kprobe handlers.
> >
> > Yes. in fact disable_trace_probe(file != NULL) does, but perf doesn't.
>
> Ah, right. we did that.

And thus we only need to synchronize kprobe_dispatcher()->kprobe_perf_func()
path. And afaics kprobe_perf_func() doesn't use anything which can be freed
by trace_remove_event_call?

> >> OTOH, unregister_kprobe() waits for that.
> >
> > Yes.
> >
> > So I think we only need to move kfree(tp->call.print_fmt).

OOPS. I was wrong. It seems that ->print_fmt is only for event/format ?
Then it is fine to kfree it right after trace_remove_event_call().

> > So the sequence should be:
> >
> > 	if (trace_remove_event_call(...))
> > 		return;
> >
> > 	/* does synchronize_sched */
> > 	unregister_kprobe();
> >
> > 	kfree(everything);
> >
> > Agreed?
>
> If we can free everything after all, I'd like to do so.
> Hmm, but AFAICS, trace_remove_event_call() supposes that
> all event is disabled completely.

Yes, but kprobe_trace_func() is really disabled?

> A safe way is to wait rcu always right after disable_*probe
> in disable_trace_probe. If we have an unused link, we can
> free it after that.

Aaaah... I am starting to understand... Even if kprobe_perf_func()
is fine, synchronize_sched() is calles _before_ disable_kprobe()
and thus it can't synchronize with the handlers which hit this probe
after we start synchronize_sched().

Did you mean this or I misssed something else?

Thanks!

Oleg.