From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752021AbWHNMjP (ORCPT ); Mon, 14 Aug 2006 08:39:15 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752033AbWHNMjP (ORCPT ); Mon, 14 Aug 2006 08:39:15 -0400 Received: from ns.suse.de ([195.135.220.2]:4795 "EHLO mx1.suse.de") by vger.kernel.org with ESMTP id S1752021AbWHNMjO (ORCPT ); Mon, 14 Aug 2006 08:39:14 -0400 From: Andi Kleen To: "Jan Beulich" Subject: Re: [PATCH for review] [127/145] i386: move kernel_thread_helper into entry.S Date: Mon, 14 Aug 2006 14:39:07 +0200 User-Agent: KMail/1.9.3 Cc: linux-kernel@vger.kernel.org References: <44E08A76.76E4.0078.0@novell.com> In-Reply-To: <44E08A76.76E4.0078.0@novell.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200608141439.07038.ak@suse.de> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Monday 14 August 2006 14:36, Jan Beulich wrote: > >> And this too - the value now in %eax has no relation with the > >> value known by the caller of this routine (which doesn't expect > >> any return from here anyway). > > > >Ok, but somehow it needs to be annotiated so that the unwinder stops > >and doesn't fall back. Can you please send a replacement patch that > >does this correctly? > > Actually, with the previous attempt we still didn't achieve the > original > goal of terminating the frame chain at kernel_thread_helper. Thus > another try (the seemingly odd extra push can be avoided once we > start using CFI expressions, which the unwinder currently doesn't > support): I added the extra push thanks (hope there weren't more changes) -Andi