From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030197AbWHOKg2 (ORCPT ); Tue, 15 Aug 2006 06:36:28 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030199AbWHOKg2 (ORCPT ); Tue, 15 Aug 2006 06:36:28 -0400 Received: from gwmail.nue.novell.com ([195.135.221.19]:16878 "EHLO emea5-mh.id5.novell.com") by vger.kernel.org with ESMTP id S1030197AbWHOKg1 (ORCPT ); Tue, 15 Aug 2006 06:36:27 -0400 Message-Id: <44E1BFDC.76E4.0078.0@novell.com> X-Mailer: Novell GroupWise Internet Agent 7.0.1 Date: Tue, 15 Aug 2006 12:36:44 +0200 From: "Jan Beulich" To: "Andi Kleen" Cc: Subject: Re: [PATCH for review] [127/145] i386: move kernel_thread_helper into entry.S Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Disposition: inline Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Just like done in the x86-64 patch that I just sent, I'd recommend moving the push added yesterday outside of the CFI-covered region (so that in the unlikely event of being caught at the push there won't be an ill assumption that a [fake] return address is already on the stack, or that there is a return address at all): ENTRY(kernel_thread_helper) pushl $0 # fake return address CFI_STARTPROC movl %edx,%eax pushl %edx CFI_ADJUST_CFA_OFFSET 4 call *%ebx pushl %eax CFI_ADJUST_CFA_OFFSET 4 call do_exit CFI_ENDPROC ENDPROC(kernel_thread_helper) Jan