From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753527AbXLEFzJ (ORCPT ); Wed, 5 Dec 2007 00:55:09 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750954AbXLEFy6 (ORCPT ); Wed, 5 Dec 2007 00:54:58 -0500 Received: from www.tglx.de ([62.245.132.106]:36345 "EHLO www.tglx.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750709AbXLEFy5 (ORCPT ); Wed, 5 Dec 2007 00:54:57 -0500 Date: Wed, 5 Dec 2007 06:54:09 +0100 (CET) From: Thomas Gleixner To: Linus Torvalds cc: Steven Rostedt , Ingo Molnar , LKML , David Holmes - Sun Microsystems , Andrew Morton , Stable Team Subject: Re: [PATCH -v2] fix for futex_wait signal stack corruption In-Reply-To: Message-ID: References: <1196801858.1645.37.camel@localhost.localdomain> <20071204213924.GA14915@goodmis.org> User-Agent: Alpine 0.99999 (LFD 796 2007-11-08) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 4 Dec 2007, Linus Torvalds wrote: > > Patch looks fine to me. > > On Tue, 4 Dec 2007, Steven Rostedt wrote: > > > > Note: I'm a bit nervious to add "linux/types.h" and use u32 and u64 > > in thread_info.h, when there's a #ifdef __KERNEL__ just below that. > > Not sure what that is there for. > > Hmm. I'd not expect user-mode headers to ever include > , and if they do, they'd already get get totally > invalid namespace pollution ("struct restart_block" at a minimum) along > with stuff that simply isn't sensible in user-space at all, so I think > this part is fine. > > And I guess somebody will scream if it bites them ;) > > Anyway, my gut feel is that this is potentially a real problem, and we > should fix it asap (ie it should go into 2.6.24 even at this late stage in > the game), but it would be nice to know if the problem actually hit any > actual real program, and not just a test-setup. > > So here's a question for David Holmes: What caused you to actually notice > this behaviour? Can this actually be seen in real life usage? > > Anyway, at a minimum, here's an > > Acked-by: Linus Torvalds > > and I suspect I should just apply it directly. Any comments from anybody > else? Doh, yes. I completely missed that stack dependency of the pointer when I looked at the patch back then. The solution looks solid and probably we should get rid of the unnamed union member and fixup the other places which use restart_block in a similar way. Just a minor nit. Can we please use "futex" instead of "fu" ? I'm just envisioning the next union member named "ba". Acked-by: Thomas Gleixner Please apply with the s/fu/futex/ change. This needs to go into stable .22/.23 as well. Thanks, tglx