From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1750924AbWDKP0U (ORCPT ); Tue, 11 Apr 2006 11:26:20 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750925AbWDKP0U (ORCPT ); Tue, 11 Apr 2006 11:26:20 -0400 Received: from e5.ny.us.ibm.com ([32.97.182.145]:914 "EHLO e5.ny.us.ibm.com") by vger.kernel.org with ESMTP id S1750914AbWDKP0T (ORCPT ); Tue, 11 Apr 2006 11:26:19 -0400 Message-ID: <443BCA98.1020805@de.ibm.com> Date: Tue, 11 Apr 2006 17:26:16 +0200 From: Carsten Otte Reply-To: carsteno@de.ibm.com Organization: IBM Deutschland User-Agent: Debian Thunderbird 1.0.7 (X11/20051017) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Linus Torvalds CC: Jes Sorensen , Andrew Morton , linux-kernel@vger.kernel.org, Hugh Dickins , Nick Piggin , bjorn_helgaas@hp.com Subject: Re: [patch] do_no_pfn handler References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Linus Torvalds wrote: > At the very least, it would also need a > > BUG_ON(is_cow_mapping(vma->vm_flags)); > > (or at least make it return VM_FAULT_SIGBUS). Because a COW mapping _will_ > confuse the VM and cause it to do random bad things in vm_normal_page(). That leaves my use-case out for now. I will need COW for my mapping when switching to this interface. Looks like a lot of things need rethinking in memory.c for COW with no struct page behind. -- Carsten Otte IBM Linux technology center ARCH=s390