From mboxrd@z Thu Jan  1 00:00:00 1970
From: Wolfgang Grandegger <wg@grandegger.com>
Subject: Re: 2.6.24-rc8-rt1: Strange latencies on mpc5200 powerpc - RCU	issue?
Date: Thu, 10 Jul 2008 09:50:47 +0200
Message-ID: <4875BF57.8030405@grandegger.com>
References: <486A3EBB.6030905@grandegger.com> <Pine.LNX.4.58.0807011036510.26650@gandalf.stny.rr.com> <486A573E.60107@grandegger.com> <486B607F.9040903@grandegger.com> <Pine.LNX.4.58.0807052041350.8466@gandalf.stny.rr.com> <48709348.6000101@grandegger.com> <20080708150826.GA30586@pengutronix.de> <4873C363.3070306@grandegger.com> <4874B4C0.5090500@pengutronix.de> <4874B9D5.1020309@grandegger.com> <20080709145217.GB30586@pengutronix.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
To: Wolfgang Grandegger <wg@grandegger.com>,
	Luotao Fu <l.fu@pengutronix.de>,
	Steven Rostedt <rostedt@goodmis.org>,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
	LKML <linux-kernel@vg
Return-path: <linux-rt-users-owner@vger.kernel.org>
Received: from mail-out.m-online.net ([212.18.0.9]:52767 "EHLO
	mail-out.m-online.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751017AbYGJHt6 (ORCPT
	<rfc822;linux-rt-users@vger.kernel.org>);
	Thu, 10 Jul 2008 03:49:58 -0400
In-Reply-To: <20080709145217.GB30586@pengutronix.de>
Sender: linux-rt-users-owner@vger.kernel.org
List-ID: <linux-rt-users.vger.kernel.org>

Luotao Fu wrote:
> Hi Wolfgang,
> On Wed, Jul 09, 2008 at 03:15:01PM +0200, Wolfgang Grandegger wrote:
>> Hi Fu (without n)
>>
> ....
>> OK, in the past you have been able to reproduce the high latencies with  
>> 2.6.24-rt1 and CONFIG_RCU_TRACE disabled, IIRC. Did you use a different  
>> toolchain at that time?
>>
> 
> Nope. As mentioned above, trace_mark() does some "real" works (what ever it is.),
> while the new mechahnismen use flags to remember the state of preemption. Maybe

I don't known what you refer to, but in __rcu_preempt_unboost() of 2.6.25.8-rt7, 
the trace code simply increments a counter:

        static void rcu_trace_boost_##type(struct rcu_boost_dat *rbd)   \
        {                                                               \
                rbd->rbs_stat_##type++;                                 \
        }

and that's the reason why latency is not affected by switching CONFIG_RCU_TRACE
on (while trace_mark uses preempt_disable/preempt_enable around).

> something here got optimized away? I take for grant, that you use gcc in your
> toolchain. Which version do you have?

The ELDK v4.2 uses:

  ppc_6xx-gcc (GCC) 4.2.2

and 

  GLIBC v2.6

But I measured the same latencies with ELDK v4.1:

  ppc_6xx-gcc (GCC) 4.0.0 (DENX ELDK 4.1 4.0.0)

  GLIBC v2.3.5

Wolfgang.