Return-Path: Received: from bombadil.infradead.org ([18.85.46.34]:36326 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751539Ab0HRU0D (ORCPT ); Wed, 18 Aug 2010 16:26:03 -0400 Date: Wed, 18 Aug 2010 16:25:59 -0400 From: Christoph Hellwig To: andros@netapp.com Cc: bhalevy@panasas.com, linux-nfs@vger.kernel.org Subject: Re: [PATCH 05/50] pnfs_submit: pnfs and nfslayoutdriver kconfig Message-ID: <20100818202559.GA16851@infradead.org> References: <1281735122-1496-1-git-send-email-andros@netapp.com> <1281735122-1496-2-git-send-email-andros@netapp.com> <1281735122-1496-3-git-send-email-andros@netapp.com> <1281735122-1496-4-git-send-email-andros@netapp.com> <1281735122-1496-5-git-send-email-andros@netapp.com> <1281735122-1496-6-git-send-email-andros@netapp.com> Content-Type: text/plain; charset=us-ascii In-Reply-To: <1281735122-1496-6-git-send-email-andros@netapp.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: MIME-Version: 1.0 On Fri, Aug 13, 2010 at 05:31:17PM -0400, andros@netapp.com wrote: > From: The pNFS Team > > Signed-off-by: Benny Halevy If you split out the Kconfig from the code it guards it should come last in the series so that the code can't be enabled until it's complete. > depends on NFS_V4 && EXPERIMENTAL > help > This option enables support for minor version 1 of the NFSv4 protocol > - (draft-ietf-nfsv4-minorversion1) in the kernel's NFS client. > + (RFC5661) including support for the parallel NFS (pNFS) features > + in the kernel's NFS client. > + > + Unless you're an NFS developer, say N. How much code does the pnfs support add to the nfs.ko module? Just including this unconditionally might not be an all that good idea. Also Trond seemed to be pretty determined to split up nfs.ko into version specific modules, and pnfs and the various layout drivers are pretty good candidates for that. Maybe the pnfs patches should be ontop of that? > +config PNFS_FILE_LAYOUT > + tristate "NFS client support for the pNFS nfs-files layout (DEVELOPER ONLY)" > + depends on NFS_FS && NFS_V4_1 is it really developers-only at this point? And if it is so why doesn't it depend on EXPERIMENTAL > + default y defaulting to y for random fringe features is frowned upon.