From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932175AbYDPOsg (ORCPT ); Wed, 16 Apr 2008 10:48:36 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1765510AbYDPOsR (ORCPT ); Wed, 16 Apr 2008 10:48:17 -0400 Received: from tomts36-srv.bellnexxia.net ([209.226.175.93]:59598 "EHLO tomts36-srv.bellnexxia.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1765397AbYDPOsQ (ORCPT ); Wed, 16 Apr 2008 10:48:16 -0400 X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: Aq8EAEKuBUhMROPA/2dsb2JhbACBYK03 Date: Wed, 16 Apr 2008 10:48:14 -0400 From: Mathieu Desnoyers To: Arjan van de Ven Cc: Ingo Molnar , Peter Zijlstra , prasad@linux.vnet.ibm.com, linux-kernel@vger.kernel.org, tglx@linutronix.de, Christoph Hellwig , "Frank Ch. Eigler" Subject: Re: [RFC PATCH 1/2] Marker probes in futex.c Message-ID: <20080416144814.GA15554@Krystal> References: <20080415123233.GA19797@Krystal> <1208264190.6395.21.camel@twins> <20080415131744.GA5248@elte.hu> <20080415134705.GB22351@Krystal> <20080415164814.GA15842@elte.hu> <20080415213832.GC7873@Krystal> <20080416131751.GI6304@elte.hu> <20080416064630.22428aea@laptopd505.fenrus.org> <20080416140009.GA13968@Krystal> <20080416072416.7b9ff3d2@laptopd505.fenrus.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline In-Reply-To: <20080416072416.7b9ff3d2@laptopd505.fenrus.org> X-Editor: vi X-Info: http://krystal.dyndns.org:8080 X-Operating-System: Linux/2.6.21.3-grsec (i686) X-Uptime: 10:46:43 up 47 days, 10:57, 4 users, load average: 0.79, 0.49, 0.39 User-Agent: Mutt/1.5.16 (2007-06-11) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * Arjan van de Ven (arjan@infradead.org) wrote: > On Wed, 16 Apr 2008 10:00:09 -0400 > Mathieu Desnoyers wrote: > > > > > > > If we want to support NMI context and have the ability to > > > > > instrument preemptable code without too much headache, we must > > > > > insure that every modification will leave the code in a > > > > > "correct" state and that we do not grow the size of any > > > > > reachable instruction. Also, we must insure gcc did not put > > > > > code between these instructions. Modifying non-relocatable > > > > > instructions would also be a pain, since we would have to deal > > > > > with instruction pointer relocation in the breakpoint code when > > > > > the code modification is being done. > > > > > > you also need to make sure no cpu is executing that code ever.. > > > but you already deal with that right? > > > > > > > By "insure that every modification will leave the code in a "correct" > > state", I mean that at any given time before, during or after the code > > modification, if an NMI comes on any CPU and try to run the modified > > code, it should have a valid version of the code to execute. Does it > > make more sense ? > > I understand your words. My concern is that I don't quite understand how you > guarantee that you'll not be executing the code you're modifying. > Just saying "it's consistent before and after" sounds nice but probably isn't > enough to be safe. > Ah, I see. I insert a breakpoint and execute a bypass rather than the code being modified. I only put back the 1st instruction byte after the rest of the instruction has been modified. > > > > Not only does the compare and jmp need to be consecutive, but the movb > > $0x0,%al also does. I *could* try to detect specific code inserted in > > between, but I really have to make sure I don't get burned by the > > compiler inserting a jmp there. > > I wonder if just sticking in 2 barriers around your code make gcc stop moving stuff too much > I'm not sure people would like the side-effect for the rest of optimizations, but it should be tried. Mathieu -- Mathieu Desnoyers Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68