From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S265163AbUGGOrH (ORCPT ); Wed, 7 Jul 2004 10:47:07 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265175AbUGGOrH (ORCPT ); Wed, 7 Jul 2004 10:47:07 -0400 Received: from sp-260-1.net4.netcentrix.net ([4.21.254.118]:10951 "EHLO asmodeus.mcnaught.org") by vger.kernel.org with ESMTP id S265163AbUGGOrE (ORCPT ); Wed, 7 Jul 2004 10:47:04 -0400 To: Christoph Pleger Cc: linux-kernel@vger.kernel.org Subject: Re: Address space for user process References: <20040707163535.02323a8f.Christoph.Pleger@uni-dortmund.de> From: Doug McNaught Date: Wed, 07 Jul 2004 10:46:56 -0400 In-Reply-To: <20040707163535.02323a8f.Christoph.Pleger@uni-dortmund.de> (Christoph Pleger's message of "Wed, 7 Jul 2004 16:35:35 +0200") Message-ID: <87hdsjswm7.fsf@asmodeus.mcnaught.org> User-Agent: Gnus/5.1002 (Gnus v5.10.2) Emacs/20.7 (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 Christoph Pleger writes: > Hello, > > I read that - at least in Kernel 2.4 - the amount of memory that can be > addressed by a user process is limited to 3 GB, no matter how much > virtual memory is present. Is it possible to raise that limit? That's limit of the ix86 architecture, not Linux. 64-bit systems running Linux have no such limit. That said, there are patches for 2.4 that create separate address spaces for kernel and user (search for '4G/4G') at the cost of a (sometimes significant) performance penalty. 4GB per-process is a hard limit on 32-bit systems, so it's the most you'll get. -Doug