From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262594AbTDQUcp (ORCPT ); Thu, 17 Apr 2003 16:32:45 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262549AbTDQUcp (ORCPT ); Thu, 17 Apr 2003 16:32:45 -0400 Received: from [12.47.58.203] ([12.47.58.203]:54227 "EHLO pao-ex01.pao.digeo.com") by vger.kernel.org with ESMTP id S262594AbTDQUcn (ORCPT ); Thu, 17 Apr 2003 16:32:43 -0400 Date: Thu, 17 Apr 2003 13:45:12 -0700 From: Andrew Morton To: Hugh Dickins Cc: rostedt@stny.rr.com, linux-kernel@vger.kernel.org, srostedt@goodmis.org Subject: Re: What's the reason that /dev/mem can't map unreserved RAM? Message-Id: <20030417134512.1623a926.akpm@digeo.com> In-Reply-To: References: X-Mailer: Sylpheed version 0.8.9 (GTK+ 1.2.10; i586-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 17 Apr 2003 20:44:34.0455 (UTC) FILETIME=[27665270:01C30522] Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hugh Dickins wrote: > > I did start out on eliminating PageReserved a few months ago, > but was persuaded to delay that until 2.7. When that's done, > you will be able to mmap /dev/mem properly. I think we can almost do it now. Just a few things like get_user_pages(), ptrace and core dumping may need to be taught about VM_RESERVED. Or we could simply change mmap_mem() to set VM_IO if any of the pages aren't valid. That's probably an accurate reflection of what the app is trying to do, too.