From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261269AbVFOWh7 (ORCPT ); Wed, 15 Jun 2005 18:37:59 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261597AbVFOWh7 (ORCPT ); Wed, 15 Jun 2005 18:37:59 -0400 Received: from smtp.osdl.org ([65.172.181.4]:23531 "EHLO smtp.osdl.org") by vger.kernel.org with ESMTP id S261269AbVFOWhy (ORCPT ); Wed, 15 Jun 2005 18:37:54 -0400 Date: Wed, 15 Jun 2005 15:39:32 -0700 (PDT) From: Linus Torvalds To: Bongani Hlope cc: Andrew Morton , ak@muc.de, Kernel Mailing List , Arjan van de Ven , Ingo Molnar Subject: Re: Tracking a bug in x86-64 In-Reply-To: <200506160020.21688.bonganilinux@mweb.co.za> Message-ID: References: <200506132259.22151.bonganilinux@mweb.co.za> <20050614132721.3b55c196.akpm@osdl.org> <200506160020.21688.bonganilinux@mweb.co.za> 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 On Thu, 16 Jun 2005, Bongani Hlope wrote: > > push 410, pop 205, pop 103, push 103, pop 51, pop 26, push 13, pop 6, push 4 and push 1 > This points to: randomisation-top-of-stack-randomization.patch Goodie. Can you verify (just to make doubly sure that there are no subtle interactions anywhere) that the current top-of-tree with that _one_ patch reverted ends up working for you? Anyway, thanks a heap for spending the time narrowing this down! Linus