From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262758AbTKZNW1 (ORCPT ); Wed, 26 Nov 2003 08:22:27 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262784AbTKZNW1 (ORCPT ); Wed, 26 Nov 2003 08:22:27 -0500 Received: from fw.osdl.org ([65.172.181.6]:5510 "EHLO mail.osdl.org") by vger.kernel.org with ESMTP id S262758AbTKZNWS (ORCPT ); Wed, 26 Nov 2003 08:22:18 -0500 Date: Wed, 26 Nov 2003 05:29:00 -0800 From: Andrew Morton To: Christoph Hellwig Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org Subject: Re: 2.6.0-test10-mm1 Message-Id: <20031126052900.17542bb3.akpm@osdl.org> In-Reply-To: <20031126130936.A5275@infradead.org> References: <20031125211518.6f656d73.akpm@osdl.org> <20031126085123.A1952@infradead.org> <20031126044251.3b8309c1.akpm@osdl.org> <20031126130936.A5275@infradead.org> X-Mailer: Sylpheed version 0.9.4 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Christoph Hellwig wrote: > > On Wed, Nov 26, 2003 at 04:42:51AM -0800, Andrew Morton wrote: > > The individual patches in the broken-out/ directory are usually > > changelogged. This one says: > > > > It was EXPORT_SYMBOL_GPL(), however IBM's GPFS is not GPL. > > > > - the GPFS team contributed to the testing and development of > > invaldiate_mmap_range(). > > > > - GPFS was developed under AIX and was ported to Linux, and hence meets > > Linus's "some binary modules are OK" exemption. > > > > - The export makes sense: clustering filesystems need it for shootdowns to > > ensure cache coherency. > > Have you actually looked at the gpfs glue code? Nope. > something that digs that deep > into the VM and VFS actually _must_ be derived work. Could be. I'm surprised that they need a glue layer at all actually. > Or do wed allow people > now to pay a developer tax to buy themselves free from GPL restrictions. Well I think that restructuring the pagecache invalidaton in such a way that it is useful for non-derived clustered filesytems does give one some rights to actually use that code. It seems a bit rude to take the code but to make it unusable. > I as one of the collective copytight holders of the kernel strongly disagree > with that, it can't be true that IBM can just ignore copyright law.. Well if people have problems with it then I don't feel strongly enough about it to dispute that, frankly. But I do not think that making a single kernel symbol inaccessible is an appropriate way of resolving a GPFS licensing dispute.