From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932423Ab1GNVvn (ORCPT ); Thu, 14 Jul 2011 17:51:43 -0400 Received: from mail-wy0-f174.google.com ([74.125.82.174]:60500 "EHLO mail-wy0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932320Ab1GNVvm (ORCPT ); Thu, 14 Jul 2011 17:51:42 -0400 Date: Thu, 14 Jul 2011 23:51:37 +0200 From: Frederic Weisbecker To: "H. Peter Anvin" Cc: Ingo Molnar , LKML , Peter Zijlstra , Jason Wessel , Thomas Gleixner Subject: Re: [PATCH 3/6] x86: Allow the user not to build hw_breakpoints Message-ID: <20110714215134.GB26786@somewhere.redhat.com> References: <1310655837-328-1-git-send-email-fweisbec@gmail.com> <1310655837-328-4-git-send-email-fweisbec@gmail.com> <4E1F5EF3.6050601@zytor.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4E1F5EF3.6050601@zytor.com> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Jul 14, 2011 at 02:26:11PM -0700, H. Peter Anvin wrote: > On 07/14/2011 08:03 AM, Frederic Weisbecker wrote: > > So that hw_breakpoints and perf can be not built on > > specific embedded systems. > > I want to emphasize I am very, very unhappy about this. It should be > possible to not build perf while still have breakpoints available... > breakpoints are way more important than perf. I know and that's a long term goal. I believe Will Deacon is currently working on breakpoints to avoid archs to translate ptrace user arch input into perf generic information (which afterward is eventually translated back to arch information). This is a necessary first step to have a ptrace breakpoint implementation independant of perf. Once we'll have that and all archs are converted to that new mode, we can work toward having breakpoints not requiring perf to work. Until we reach that point, it's still nice to have optable breakpoints. I think it is a desired feature given the feedbacks I get from Prasad for example. If that makes it possible to try a kernel without perf in x86, good too. But this is not intended to be a solution of the breakpoint - perf dependency. That dependency was a design mistake that needs more to be fixed. We want to be able to have breakpoints without perf in the end. This just requires deeper and longer term work that is currently in progress.