From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261253AbUDBWmx (ORCPT ); Fri, 2 Apr 2004 17:42:53 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261258AbUDBWmw (ORCPT ); Fri, 2 Apr 2004 17:42:52 -0500 Received: from newport.ucsd.edu ([132.239.73.89]:59795 "EHLO localhost.localdomain") by vger.kernel.org with ESMTP id S261253AbUDBWmu (ORCPT ); Fri, 2 Apr 2004 17:42:50 -0500 Message-ID: <406DEC69.6010909@physics.ucsd.edu> Date: Fri, 02 Apr 2004 14:42:49 -0800 From: Terrence Martin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040124 X-Accept-Language: en-us, en MIME-Version: 1.0 To: linux-kernel@vger.kernel.org Subject: Bigphysarea patch and 2.6, advice requested Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi, I have been asked to investigate the 2.6 kernel for a high performance computing application. One of the questions we have is whether or not 2.6 supports the so called "bigphysarea" patch, or "hack" as it is sometimes referred to or has some built in mechanism to accomplish the same ends, large contiguous blocks of memory suitable for DMA. The problem is that the project would like to use large contiguous areas of memory for the purposes of DMA for various custom made PCI cards. I have read in a couple of posts that when the PCI device supports scatter/gather that is the preferred mechanism. The issue is that there is a very good chance that not all of the devices have the scatter/gather mechanism implemented on board. Is there anything in 2.6 that accomplishes the same ends as the 2.4 based bigphysarea patch/hack? What are other people doing when they need this functionality and also want to run 2.6? Thanks for any insight you can provide. Terrence Martin UCSD Physics