Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1161213AbWJDPMZ (ORCPT ); Wed, 4 Oct 2006 11:12:25 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1161180AbWJDPMZ (ORCPT ); Wed, 4 Oct 2006 11:12:25 -0400 Received: from smtp.osdl.org ([65.172.181.4]:25828 "EHLO smtp.osdl.org") by vger.kernel.org with ESMTP id S1161213AbWJDPMY (ORCPT ); Wed, 4 Oct 2006 11:12:24 -0400 Date: Wed, 4 Oct 2006 08:12:10 -0700 (PDT) From: Linus Torvalds To: Jan Beulich cc: Ingo Molnar , Eric Rannaud , Andrew Morton , Andi Kleen , Chandra Seetharaman , Linux Kernel Mailing List , nagar@watson.ibm.com Subject: Re: BUG-lockdep and freeze (was: Arrr! Linux 2.6.18) In-Reply-To: <45239953.76E4.0078.0@novell.com> Message-ID: References: <5f3c152b0609301220p7a487c7dw456d007298578cd7@mail.gmail.com> <200609302230.24070.ak@suse.de> <20060930204900.GA576@elte.hu> <45239953.76E4.0078.0@novell.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1401 Lines: 39 On Wed, 4 Oct 2006, Jan Beulich wrote: > > Again, a corrupt stack will not allow you getting reliable data out of the > old unwinder either. Even worse when you consider a stack overflow and > your request for range checks (or pointers into the stack) - you might not > get a stack trace then at all. Who cares? Not getting data out of a corrupt stack isn't the problem. Getting an OOPS is the problem. Jan, did you follow the actual thread at all? This is _not_ about getting "perfect data". This is about a dead machine that was killed not by the original bug, but by the DEBUGGING CODE it triggered. In other words, in this case the debugging code made things harder to debug. That's bad, bad, bad. That negates the whole point of having debugging code in the first place. In contrast, the old x86 (32-bit) stack dumper was damn safe. If the original %esp was at all a valid pointer (and it had to be, since otherwise you'd have gotten a double fault!), it would print out something _without_ crapping all over the machine. Maybe it wouldn't print out anything nice, but it would never make things WORSE. See the difference? Linus - 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/