From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754118Ab3KUPU5 (ORCPT ); Thu, 21 Nov 2013 10:20:57 -0500 Received: from merlin.infradead.org ([205.233.59.134]:50183 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753876Ab3KUPU4 (ORCPT ); Thu, 21 Nov 2013 10:20:56 -0500 Date: Thu, 21 Nov 2013 12:12:58 -0300 From: Arnaldo Carvalho de Melo To: Steven Rostedt Cc: Borislav Petkov , Ingo Molnar , LKML , Borislav Petkov , Jiri Olsa , Peter Zijlstra , Robert Richter Subject: Re: [PATCH] perf: Move fs.* to generic lib/lk/ Message-ID: <20131121151258.GB24806@ghostprotocols.net> References: <1384984585-24300-1-git-send-email-bp@alien8.de> <20131121073410.GA16971@gmail.com> <20131121100722.GA25459@pd.tnic> <20131121111744.GA27513@gmail.com> <20131121113014.GA26009@pd.tnic> <20131121114224.GA27704@gmail.com> <20131121120605.GC26009@pd.tnic> <20131121123954.GA8044@home.goodmis.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20131121123954.GA8044@home.goodmis.org> X-Url: http://acmel.wordpress.com User-Agent: Mutt/1.5.21 (2010-09-15) X-SRS-Rewrite: SMTP reverse-path rewritten from by merlin.infradead.org See http://www.infradead.org/rpr.html Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Em Thu, Nov 21, 2013 at 07:39:55AM -0500, Steven Rostedt escreveu: > On Thu, Nov 21, 2013 at 01:06:05PM +0100, Borislav Petkov wrote: > > No, I absolutely don't think it is a bikeshedding exercise too. We > > really need that lib because people have started growing their own > > facilities in tools/ and outside so having a lib which implements > > generic functionality for interfacing with the kernel right is going to > > save everyone a lot of time and energy. > Should we put libtraceevent here too? Have all kapi functionality live here? Don't know, what I'm liking about this discussion now is that we're focusing on breaking it down into multiple libraries. So at some point, symbol resolution, that involves parsing multiple "fs based kernel api" (/proc/{kallsyms,modules,etc), but should not be in the library being discussed now, will be another tools/perf/util/ "spin-off", that, again, focuses on a clear mission statement and is useful for multiple /usr/src/linux/tools/ "clients". And as the first step, only tools/ clients would be targeted, as we would have some time to do changes that touches the libraries _and_ all its users, just like we can do with kernel APIs and its users, i.e. drivers, etc. Only at a later stage, when those libraries are boringly mature, we would export them for wider use. > As libtraceevent is getting mature, we need to get it out as a real library > soon. There is already three utilities that use it: perf, trace-cmd and > powertop. And I've heard of others that want that functionatity. > > -- Steve