From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262151AbTEEMC7 (ORCPT ); Mon, 5 May 2003 08:02:59 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262155AbTEEMC7 (ORCPT ); Mon, 5 May 2003 08:02:59 -0400 Received: from pc2-cwma1-4-cust86.swan.cable.ntl.com ([213.105.254.86]:43172 "EHLO lxorguk.ukuu.org.uk") by vger.kernel.org with ESMTP id S262151AbTEEMC6 (ORCPT ); Mon, 5 May 2003 08:02:58 -0400 Subject: Re: The disappearing sys_call_table export. From: Alan Cox To: Terje Eggestad Cc: Christoph Hellwig , Arjan van de Ven , Linux Kernel Mailing List , D.A.Fedorov@inp.nsk.su In-Reply-To: <1052127216.2821.51.camel@pc-16.office.scali.no> References: <1052122784.2821.4.camel@pc-16.office.scali.no> <20030505092324.A13336@infradead.org> <1052127216.2821.51.camel@pc-16.office.scali.no> Content-Type: text/plain Content-Transfer-Encoding: 7bit Organization: Message-Id: <1052133402.29361.2.camel@dhcp22.swansea.linux.org.uk> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 05 May 2003 12:16:43 +0100 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Llu, 2003-05-05 at 10:33, Terje Eggestad wrote: > 1. performance is everything. Then you can live with building custom patched kernels > 2. We're making a MPI library, and as such we don't have any control > with the application. LD_PRELOAD > 3c. It's therefore necessary for HW to access user pages. Like TV cards do. That isnt hard > 4. In order to to 3, the user pages must be pinned down. > 5. the way MPI is written, it's not using a special malloc() to allocate > the send receive buffers. It can't since it would break language binding > to fortran. Thus ANY writeable user page may be used. Well not all the pages are guaranteed DMAable, so I guess you already lost. > 10. kernel patches are impractical, I must be able to do this with std > stock, redhat, AND suse kernels. So you want every vendor to screw up their kernels and the base kernel for an obscure (but fun) corner case. Thats not a rational choice is it. You want "performance is everything" you pay the price, don't make everyone suffer.