From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Date: Tue, 16 Nov 2010 09:54:56 +0100
From: Martin Schwidefsky <schwidefsky@de.ibm.com>
Subject: Re: [RFC][PATCH v2 5/7] taskstats: Improve cumulative CPU time
 accounting
Message-ID: <20101116095456.7389bebc@mschwide.boeblingen.de.ibm.com>
In-Reply-To: <1289844606.2109.526.camel@laptop>
References: <20101111170352.732381138@linux.vnet.ibm.com>
	<20101111170815.404670062@linux.vnet.ibm.com>
	<20101113183810.GA9021@redhat.com>
	<20101115165521.21baac60@mschwide.boeblingen.de.ibm.com>
	<1289837036.2109.501.camel@laptop>
	<20101115184910.549a828b@mschwide.boeblingen.de.ibm.com>
	<1289843517.2109.521.camel@laptop>
	<20101115190031.5a50e760@mschwide.boeblingen.de.ibm.com>
	<1289844606.2109.526.camel@laptop>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-Archive: <https://lore.kernel.org/lkml/>
List-Post: <mailto:linux-kernel@vger.kernel.org>
To: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>, Michael Holzheu <holzheu@linux.vnet.ibm.com>, Shailabh Nagar <nagar1234@in.ibm.com>, Andrew Morton <akpm@linux-foundation.org>, Venkatesh Pallipadi <venki@google.com>, Suresh Siddha <suresh.b.siddha@intel.com>, Ingo Molnar <mingo@elte.hu>, John stultz <johnstul@us.ibm.com>, Thomas Gleixner <tglx@linutronix.de>, Balbir Singh <balbir@linux.vnet.ibm.com>, Heiko Carstens <heiko.carstens@de.ibm.com>, Roland McGrath <roland@redhat.com>, linux-kernel@vger.kernel.org, linux-s390@vger.kernel.org
List-ID: <linux-s390.vger.kernel.org>

On Mon, 15 Nov 2010 19:10:06 +0100
Peter Zijlstra <a.p.zijlstra@chello.nl> wrote:

> On Mon, 2010-11-15 at 19:00 +0100, Martin Schwidefsky wrote:
> > On Mon, 15 Nov 2010 18:51:57 +0100
> > Peter Zijlstra <a.p.zijlstra@chello.nl> wrote:
> > 
> > > On Mon, 2010-11-15 at 18:49 +0100, Martin Schwidefsky wrote:
> > > > > To what purpose? 
> > > >  
> > > > Is that a trick question? Why do we have tools like "top"? Or process
> > > > accounting? The point is that the quality of the numbers we get right
> > > > now is rather bad, the overhead of scanning /proc is horrendous and
> > > > the 10ms granularity is rather coarse. 
> > > 
> > > But you're not just replacing top, you're adding all kinds of new
> > > accounting crap all over the place.
> > 
> > We DO replace top. Patch #7 of 7.
> 
> You _also_ replace top, but its not by far the only thing you do. If you
> simply replaced top you wouldn't need to add tons of extra accounting,
> would you?

There are basically two things we want to accomplish:
1) Make top faster by replacing the /proc based data gathering with taskstats.
   To really make it go fast with taskstats we filter with last_depart to
   read only tasks that have changed since the last snapshot.
2) Make top more precise. That is where all of the extra accounting comes
   into play.

-- 
blue skies,
   Martin.

"Reality continues to ruin my life." - Calvin.