From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S265489AbUHFMwl (ORCPT ); Fri, 6 Aug 2004 08:52:41 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265897AbUHFMwl (ORCPT ); Fri, 6 Aug 2004 08:52:41 -0400 Received: from science.horizon.com ([192.35.100.1]:64807 "HELO science.horizon.com") by vger.kernel.org with SMTP id S265489AbUHFMwh (ORCPT ); Fri, 6 Aug 2004 08:52:37 -0400 Date: 6 Aug 2004 12:52:36 -0000 Message-ID: <20040806125236.24348.qmail@science.horizon.com> From: linux@horizon.com To: mbligh@aracnet.com Subject: Re: HIGHMEM4G config for 1GB RAM on desktop? Cc: linux-kernel@vger.kernel.org Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org [Resend because my fingers STILL type "@vger.rutgers.edu" if I'm not paying attention...] > In practice, I suspect 2/2 will do exactly what you want ... and what > 99.9% of people want actually ;-) We could add more options, but be sure > to mark anything that's not 1GB aligned as not suitable for PAE (as the > 0.5 split was). But if you're using PAE, you've got > 4G of RAM, so there's no need to be clever trying to avoid HIGHMEM options. Unfortunately, I just had a server with Con's patch detonate overnight (Oops in interrupt -> panic; details in separate e-mail), and wli tells me that there are additional places in the code that need fixing. I notice that all previous patches had the kernel range a power of 2 in size. Is this required somewhere? I thought it was just that the kernel had to start at a PGD boundary (4M on normal x86, 1G on PAE). If 128M is always enough, a split at 0xb800000 seems possible, but giving it an extra 128M seems like a nice bit of safety for PCI devices.