From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752768AbXCSTkO (ORCPT ); Mon, 19 Mar 2007 15:40:14 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752757AbXCSTkO (ORCPT ); Mon, 19 Mar 2007 15:40:14 -0400 Received: from ebiederm.dsl.xmission.com ([166.70.28.69]:39350 "EHLO ebiederm.dsl.xmission.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752768AbXCSTkM (ORCPT ); Mon, 19 Mar 2007 15:40:12 -0400 From: ebiederm@xmission.com (Eric W. Biederman) To: "H. Peter Anvin" Cc: Avi Kivity , David Miller , dmlb2000@gmail.com, linux-kernel@vger.kernel.org Subject: Re: PAGE_SIZE Availability Inconsistency References: <9c21eeae0703051555x1884fd7cse7968a71ec04eb27@mail.gmail.com> <20070305.160303.59493440.davem@davemloft.net> <9c21eeae0703051604l20a029c3r8cf09d792ba523be@mail.gmail.com> <20070305.162620.04006691.davem@davemloft.net> <45ECD03F.2040008@zytor.com> <45F07B63.1000105@argo.co.il> <45F08C60.6010404@zytor.com> Date: Mon, 19 Mar 2007 13:39:29 -0600 In-Reply-To: <45F08C60.6010404@zytor.com> (H. Peter Anvin's message of "Thu, 08 Mar 2007 14:21:20 -0800") Message-ID: User-Agent: Gnus/5.110006 (No Gnus v0.6) Emacs/21.4 (gnu/linux) 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 "H. Peter Anvin" writes: > Avi Kivity wrote: >>> >>> It definitely should, especially on x86-64, where the page size isn't >>> guaranteed by the ABI (on i386, the ABI guarantees a 4K page size; on x86-64 >>> it can be up to 64K.) >> >> Wouldn't that be ia64? > > No, the x86-64 EFI ABI permits page sizes up to 64K. Currently, of course, the > only page size in use is 4K, but unlike i386 that's not guaranteed by the ABI. > At least AMD has indicated that they are considering introducing larger page > size support in future hardware. EFI ABI? Don't you mean the SYSV ABI right. That does seem to indicate that 64K is an option. And at a quick glance glibc is using 1M alignment on program segments so it looks like there is at least a reasonable chance of being able to make the transition to a bigger page size for user space. Eric