Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1762858AbYBLK3l (ORCPT ); Tue, 12 Feb 2008 05:29:41 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1761653AbYBLK3b (ORCPT ); Tue, 12 Feb 2008 05:29:31 -0500 Received: from mx1.redhat.com ([66.187.233.31]:39597 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1761395AbYBLK3a (ORCPT ); Tue, 12 Feb 2008 05:29:30 -0500 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: Roland McGrath To: Andi Kleen X-Fcc: ~/Mail/linus Cc: Ingo Molnar , "Frank Ch. Eigler" , linux-kernel@vger.kernel.org, Thomas Gleixner , "H. Peter Anvin" , Linus Torvalds , Andrew Morton Subject: Re: [git pull] kgdb-light -v9 In-Reply-To: Andi Kleen's message of Tuesday, 12 February 2008 11:03:27 +0100 <20080212100327.GA30873@one.firstfloor.org> References: <20080211015321.GA27376@one.firstfloor.org> <20080211162141.GA31434@elte.hu> <20080211171039.GA20446@one.firstfloor.org> <20080211230335.GA16102@elte.hu> <20080212100327.GA30873@one.firstfloor.org> X-Windows: form follows malfunction. Message-Id: <20080212102636.02715270192@magilla.localdomain> Date: Tue, 12 Feb 2008 02:26:35 -0800 (PST) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1169 Lines: 24 > You silently overwrite any user ptrace hw breakpoints right? To do it cleanly > would still require a reservation frame work. There was work underway on that before (hw_breakpoint). I'm not entirely sure you want to use fancy stuff like that in kgdb. It's nice for kgdb to be as self-contained as possible, so you can debug everything else with it. If you're going to rely on lots of higher-level infrastructure, it could be using kprobes for its software breakpoint handling, too. At any rate, I think it would be good if the hw breakpoint support in kgdb were chopped out into a separate patch. First make kgdb work with no code touching debug registers at all. Then a second patch can add the hw breakpoint support. The latter one can be mulled over or replaced with a hw_breakpoint merge or just put off for a while because it's a big can of worms to multiplex the use of the debug registers. Thanks, Roland -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/